November 16, 2026

Introducing spatial CSS: A developer preview

Surfin’ Safari

Safari on visionOS 27.3 comes with an important update: a limited feature preview of the proposed spatial CSS API, with an initial focus on working with 3D objects through the <model> element. While it’s a big conceptual jump, spatial CSS itself is made of a lot of small tweaks to existing syntax. Read on for more details, but here’s a list of each of the features covered in the update:

  • Collecting multiple elements inside a spatial: portal
  • transform to move multiple models independently
  • position-anchor to snap models together
  • @environment-map syntax to reuse and manage different lighting,
  • Enabling orbit over the portal scene with portal-action:orbit,
  • Using requestVolumetricScene() to launch a connected volumetric scene.

To enable the features for yourself on Vision Pro, or to check it out in the visionOS Simulator, go to
Settings > Apps > Safari > Advanced > Feature flags > CSS Portals.
You can experience the spatial CSS demo page here. You can also learn more about installing additional simulators from Xcode here.

The big picture

The 3D <model> element, now available in Safari 27 on all platforms, allows you to present intrinsically 3d objects on their own, in their own space. However, the web is at its best when it allows the composition and configuration of content based on author intentions and user context. These context cues can include the device display capabilities, window size, and color theme. These requirements will remain even as we include spatial content — and ideally, authors should be able to use the same kinds of patterns to build for the spatial web.

And spatial content should not just be designed for spatial devices. The last few years have seen a dramatic rise in the number of websites showing 3D content – on phones, tablets and computers as well as headsets. Spatial web standards should support designing for browsers wherever they run.

New web features, classical CSS

By routing functionality for spatial content through declarative CSS, a page can use media queries targeting aspect ratio, color theme and reduced motion to adapt spatially as well. This adds much more flexibility to your website without having to do as much in your JavaScript, and removing the need to keep track of any complicated state.

Even better, the numerical parameters can be animated and transitioned in the same way that any other value can be changed on the page.

visionOS 27.3 offers the first components of spatial CSS in a developer preview, and this article demonstrates how to use them through the case study of a simple set of children’s blocks.

Feature breakdown

spatial: portal

The feature that kicks this all off is the declaration that an element is acting as a portal, through the new spatial:portal syntax. Elements that can contain other elements – think <figure>, <div> – can act as a portal, and will then act as a space that all of the child elements nested inside it share. This preview focuses on model elements, but the specification is written so that this approach should work for all content.

spatial transform:

Once the models are collected inside a shared space, you can then move, rotate, and scale them using their standard transform attribute. This takes the same DOMMatrix class that creates 2D and pseudo-3D transforms in other contexts, but can now act as a true 3D transform to objects inside the portal context.

A blue shape-sorter puzzle board on a white background with four recesses: an octagon, a star, a cross and a triangle. An orange octagon block sits to the left of the board and an orange star block to the right."
A simple scene of shaped blocks and a board to slot them in.
<div class="world" style="spatial:portal;">
  <model id="board" src="board.usdz"></model>
  <model id="octagon" src="octagon.usdz"></model>
  <model id="star" src="star.usdz"></model> 
</div>
<style>
    #octagon { 
        transform:translateX(-50cm);
    }
    #star { 
        transform:translateX(50cm);
    }
</stye>

The code for presenting this scene and setting the blocks off to the sides.

The board and the block objects can be moved and rotated independently — an expanded settings panel shows different values. While a model’s animation timeline could manage single sequence of action, this adds the degrees of freedom required for more sophisticated interaction or configuration.

position-anchor

Fitting 3D objects together can get complicated. We need a way to attach models to specific places on other models — think of putting together pieces of furniture. Model is a void element, which means it means we can’t nest models inside other model elements the same way that other DOM is nested. Instead, we’re borrowing from the CSS concept of position-anchor and adding a way to align and orient objects according to their anchors. position-context: anchor is the way to make it work.

See it in action

The blue shape-sorter board on a light gray background with the orange octagon block fitted into its recess. The star, cross and triangle recesses are still empty, and the orange star block sits loose to the right of the board.
The scene presented earlier with the octagonal block slotted exactly into place.
<div class="world">
    <model id="board" src="board.usdz"></model>
    <model id="octagon" src="octagon.usdz"></model>
    <model id="star" src="star.usdz"></model>
</div>

<style>
    #board {
        anchor-name:--board;
    }
    #octagon {
        position-anchor:--board#octagon-position;
        position-context:anchor;
    }    
    #star {
        display:none;
    }
</style>

The code for using position-anchor to achieve the correct placement for the block.

The board used in the example consists of three separate objects inside the 3D program it was built within: The visible shape of the board itself, and two reference positions — often called “null objects” or “empties” in 3D editors. These correspond to the correct position for each of the shape blocks to slot into when they’re on the board. Here’s the scene and its contents inside Blender:

A screenshot of the Blender 5.3 Beta interface with shapeblocks.blend open. The 3D viewport shows the blue shape-sorter board and the orange octagon and star blocks. The outliner on the right lists block-star, block-octagon, target-octagon, target-star and board, and the timeline is at frame 168.
The default view of the block scene presented inside Blender. Note both the block objects and the target objects in-place on the board.

Rather than copy-pasting values from your 3D editor, or having to guess at the right transform, you can nominate specific anchor points — just like in the 2D equivalent of the feature — to place an object in the right location. These anchor points can be animated, too, making it easier to compose elements even when they’re moving around.

@environment-map

An environment map is the lighting that is reflected off of an object — think of it as the room that an object appears within.

We introduced environment maps as an HTML attribute on the model element, because it was the only element that would be able to use it. Now that there is a broader range of elements that will be able to share environment maps. Taking inspiration from @keyframes for animation, we leverage @environment-map to facilitate this. This preview provides support for 360º HDR images and adds the clarifying syntax to indicate what they are and how to work with them.

See it in action

"The shape-sorter board and both blocks in a dark scene lit from below. The board looks near-black against a murky brown background. Green, purple and orange light glows in the recesses and along the edges of the orange octagon and star blocks.
The scene showing a spooky-colored underlighting effect.
<div class="world spooky">
    <!--... -->
</div>
@environment-map {
    name: "spooky";
    format: equirectangular;
    src: url("night.hdr");
}

.world.spooky {
     environment-map: "spooky";
}

The code for setting up the spooky environment lighting.

portal-action:orbit

One of the main things that users want to do with 3D content is to look around it by rotating the scene in-place. In 3D programs this is often called an ‘orbit’, and spatial CSS enables it with the new portal-action attribute. It may be useful to add more modes in the future like pan-x or pan-y. This mirrors the model element’s stagemode attribute on the HTML element, but by configuring it in CSS it becomes more flexible.

requestVolumetricScene()

In addition to showing a spatial arrangement inside a portal, visionOS has the ability to show these scenes outside of the browser window in what we’re calling a Connected Volumetric Scene. Launch a request from a user action to remove the spatial content from its portal and present it in the user’s space:

launchSpaceButton.addEventListener("click", event=>{
    const world = document.querySelector(".world");
    world.requestVolumetricScene();
});

The code to launch the volumetric scene.
Users can place the volumetric scene anywhere in their space, and the live link between the web page and the volumetric scene will let you update the scene — add or remove objects, or move them based on your needs.

Your feedback is essential

There is a long journey ahead, and a lot more to get to in the proposed spatial CSS specification. While this feature preview is limited to visionOS, we thought it was important to release along with the full specification as an indication of how to approach bringing spatial concepts to the web. Along with the general release of the <model> element on Safari 27, we hope this gives authors enough to think about the opportunities — and requirements — that lie ahead in the spatial web. That’s also why it’s so important for authors to explore the feature preview and tell us about their experience of using it. You can download the visionOS simulator with Xcode. If you don’t already have Xcode, you can download Xcode from the Mac app store.

To share your thoughts on the Spatial CSS feature preview, find us on Mastodon at @jensimmons@front-end.social, @jondavis@mastodon.social, or on X at @zachernuk. Or send a reply on X to @webkit. You can also follow WebKit on LinkedIn. If you run into any issues, we welcome your feedback about WebKit at bugs.webkit.org. Filing issues really does make a difference. Thanks.

November 16, 2026 11:45 PM

Igalia WebKit Team: WebKit Igalia Periodical #86

Igalia WebKit

Update on what happened in WebKit in the week from November 9 to November 16.

The bag of updates this week is composed of improvements to click events generation, improved scale factor and fixed rendering of SVG images, and the introduction of the new ATrace backend for performance analysis on WPE Android.

Cross-Port 🐱

Taps now correctly generate click events when touchmove is prevented, as long as movement stays within the tap threshold.

Graphics 🖼️

Made SVG images use the device scale factor of the document drawing them, instead of always rendering at 1x. On high-DPI displays, resolution media queries inside the image now match the real display, and HTML content in <foreignObject> is now aligned to the display's actual pixels and its CSS filters render at full resolution.

Fixed masks and patterns going stale after their element is scaled or resized in the Layer-Based SVG Engine (LBSE). Their cached images are now kept per element and scale and are dropped when the element's bounds change, so an element painted at several scales (as inside nested patterns) gets a correctly sized image for each scale.

WPE WebKit 📟

WPE Android ↗ 🤖

Adaptation of WPE WebKit targeting the Android operating system.

Added an ATrace backend for performance analysis on WPE Android. Until now the port built with Sysprof capture disabled, so every WTF::tracePoint() and signpost compiled to nothing; now they land in the same systrace/Perfetto timeline as SurfaceFlinger, input dispatch, and the scheduler. The compositor's FPS counter is also computed while ATrace is on. The backend is selected with the new USE_ATRACE_CAPTURE build option, which is on by default when building for Android and replaces Sysprof capture there.

That’s all for this week!

By Igalia WebKit Team at November 16, 2026 07:04 PM

November 13, 2026

Safari on Apple HomeView

Surfin’ Safari

It’s not every day our browser begins a life on a whole new platform. Now, Safari also lives on Apple HomeView, joining Safari on macOS, iOS, iPadOS, and visionOS.

HomeView with a hand reaching out to touch the screen, with a Safari app icon showing in the middle

People interact with HomeView with their voice from across the room, or up close with a tap on the square, 7-inch display. With easy access to the web in Safari, users can quickly look something up, open a link from a friend, cook from a recipe, or much more.

As a web developer, you might be wondering “What do I need to do to support Safari on Apple HomeView?” Most likely, you don’t need to do anything. This is Safari, powered by WebKit. And the web itself is designed to work on any screen, with any kind of input.

However, there might be a few things you can do to make your site shine in Safari on HomeView.

With the new cooking experience, Safari recognizes when users are on a webpage with a recipe, and offers a streamlined view of step-by-step instructions.

The cooking experience in Safari on Apple HomeView. A photo of Butternut Squash Soup fills the screen. Two buttons towards the bottom offer options for seeing ingredients or start cooking.

Safari creates this view from recipe data in the schema.org recipe type in JSON-LD. If you have recipes on your site, and don’t yet support this schema.org type, mark up the recipe details in JSON and add it to the page.

<script type="application/ld+json">
{
    "@context": "[https://schema.org](https://schema.org/)",
    "@type": "Recipe",
    "name": "Perfect Roast Chicken",
    "image": "https://example.com/roast-chicken.jpg",
    "prepTime": "PT15M",
    "cookTime": "PT1H20M",
    "recipeYield": "4 servings",
    "recipeIngredient": [
        "1 whole chicken (4–5 lb)",
        "3 tbsp unsalted butter, softened"
    ],
    "recipeInstructions": [
        { "@type": "HowToStep", "text": "Preheat the oven to 425°F (220°C)." },
        { "@type": "HowToStep", "text": "Rub the butter all over the skin." }
    ]
}
</script>

HomeView also provides rich experiences playing audio and video. To ensure great experiences with video captions, include them as WebVTT files inside track elements.

<video controls>
  <source src="theshow.mp4" type="video/mp4">
  <source src="theshow.webm" type="video/webm">
  <track default
    kind="captions"
    srclang="en"
    label="English"
    src="captions.vtt">
</video>

Sites built on modern best practices work well on HomeView. Sites that hard-code assumptions about what a device is or how people use it can break when a new device arrives. A few of those assumptions are worth checking.

Sometimes sites assume every user clicks or taps, when many people use other input methods instead. On HomeView, voice is a key part of how people use the device. Someone speaking to a page isn’t pointing at a button. They are calling its name. Semantic HTML makes that work, with real buttons, links, and form fields, and a clear label on every control. Accessible content, large tap targets, and flexible layouts help everyone use your site, whether they speak, touch, or both.

Some sites assume an iOS user agent means the device is an iPhone. Yet, Safari on HomeView also reports an iOS user agent. User agent sniffing can result in unintended consequences as technology evolves, so it’s best to avoid it.

  • To find out whether an API is supported, use feature detection instead and test directly for support.
  • Instead of using UA sniffing to present a custom banner encouraging users to download your app, use Smart App Banners. Such a banner might not make sense on HomeView, and you can rely on Safari to figure that out.

You’ll end up with a far more robust website, prepared for whatever the future brings.

Some sites assume a screen is either landscape or portrait, and never square. Some sites assume the viewport is always taller than 600px, while on HomeView it’s shorter. Such assumptions can cause layout problems. Sticky or fixed-position headers and footers might leave almost no room for the main content on a short screen. Modals (newsletter signups, etc.) taller than the viewport might leave users with no way to dismiss them. One quick fix for such modals is layout code like this:

.modal {
  max-height: 90svh;
  overflow-y: auto;
}

If you don’t have HomeView handy, test your site on a phone in landscape orientation, and solve any problems that emerge due to the short height. You can also use Responsive Design Mode in Safari, and set the viewport to 448×364.

Most likely, your website already works great in Safari on HomeView — sitting on someone’s kitchen counter or hung on the wall in their hallway. That’s the promise of the web. Leverage modern best practices to build robust, flexible websites — and your work will reach people in places you might not have imagined.

November 13, 2026 09:00 PM

November 10, 2026

Igalia WebKit Team: WebKit Igalia Periodical #85

Igalia WebKit

Update on what happened in WebKit in the week from November 2 to November 10.

This week it's not all fun and games, yet gamepad support gets some polish with reduced latency and support for analog buttons; WebAudio contexts now support choosing outputs; the old compositor gets removed; and another flurry of small changes make their way to the new SVG rendering engine.

WebKitGTK 🖥️

Fixed incorrect gamepad button mapping in WebKitGTK, where the North and West facing direction buttons were swapped.

Improved gamepad input latency, reducing delays from up to 50ms down to the 10 to 12ms range, and preventing missed quick button presses.

Cross-Port 🐱

Added support for analog gamepad buttons when using libmanette.

Fixed a bug that caused formAssociatedCallback to sometimes fire twice in Safari.

Multimedia 🎥

GStreamer-based multimedia support for WebKit, including (but not limited to) playback, capture, WebAudio, WebCodecs, and WebRTC.

Implemented support for the WebAudio AudioContext.setSinkId() API in the GStreamer multimedia backend.

Graphics 🖼️

Since shipping the new Skia-based compositor enabled by default in September, the team has shipped fixes for a few issues in the 2.54.1 and 2.54.2 releases. The Skia compositor now supports more features than the TextureMapper compositor did, while reducing memory usage and providing better performance.

At this point keeping the TextureMapper code in the source tree did no longer provide any benefit and it was not actively tested, so it was decided to no longer support building TextureMapper to ease maintenance going forward. It has served WebKit well for 16 years.

Fixed a crash in the Layer-Based SVG Engine (LBSE) that happened when a shape filled or stroked with an SVG <pattern> got its renderer before the pattern existed. The pattern's image cache now drops a destroyed shape's entry by itself, and the shape is registered as a client once the pattern appears, so no stale cache entries are left behind.

Fixed the composited bounds of accelerated filter animations, which ignored the blur outsets of the animated filter because animations running on the compositor don't trigger style updates. WebKit now uses the largest outsets across all filter key frames for overlap testing and view transition captures, so a later sibling that only overlaps the blurred area is composited correctly and no longer painted below the animated layer

Made SVG fonts obey the font-variant attribute of <font-face>, so a face now only matches the variants it lists, following CSS 2 font matching. Text with small-caps now uses the glyphs of a small-caps face directly instead of synthesizing scaled-down capitals that these fonts often lack, and normal text no longer depends on the order in which the faces were registered.

WPE WebKit 📟

WPE Platform API 🧩

New, modern platform API that supersedes usage of libwpe and WPE backends.

Fixed named cursor images being drawn away from the actual pointer location when using the built-in WPEPlatform Wayland implementation.

Infrastructure 🏗️

Replaced the frame cycle with composition-aware reporting in webkit-sysprof, the tool that analyzes Sysprof captures from WebKitGTK and WPE, so frames that the ThreadedCompositor produces on its own (such as for accelerated animations) are no longer missed. A new timeline command draws the main thread and the compositor as “swimlanes” with numbered frames, which makes dropped frames much easier to track down.

That’s all for this week!

By Igalia WebKit Team at November 10, 2026 11:04 PM

November 05, 2026

Release Notes for Safari Technology Preview 256

Surfin’ Safari

Safari Technology Preview Release 256 is now available for download for macOS Golden Gate and macOS Tahoe. If you already have Safari Technology Preview installed, you can update it in System Settings under General → Software Update.

This release includes WebKit changes between: 323398@main…324653@main.

Accessibility

Resolved Issues

  • Fixed list marker text generated with the content property not being exposed in the accessibility tree. (323578@main) (189838099)
  • Fixed VoiceOver being unable to reach the content of an iframe nested inside an iframe with role="none" or role="presentation". (323749@main) (189901948)

Animations

Resolved Issues

  • Fixed scroll-driven animations drawing a translate keyframe of none as the element’s own translate value. (323583@main) (189882533)
  • Fixed: Improved performance of scroll-driven animations by avoiding style invalidation on animation updates when timing is unchanged. (324449@main) (190709310)

CSS

New Features

  • Added support for text-combine-upright on ::marker. (323582@main) (189912173)
  • Added support for continue: collapse with line-clamp, which is now set by the line-clamp shorthand. (323734@main) (190023170)
  • Added support for font-palette-mix() on Cocoa platforms, allowing font color palettes to be blended. (324044@main) (190133191)

Resolved Issues

  • Fixed ::highlight() styles not being found past anonymous boxes. (323588@main) (148245032)
  • Fixed text zoom being applied twice to font sizes specified with font-relative units. (324195@main) (188215447)
  • Fixed aspect-ratio interpolation being inverted in vertical writing modes. (323584@main) (188500448)
  • Fixed document.fonts.ready never resolving when an unused @font-face has a failing local() source followed by a url() source. (323883@main) (189243828)
  • Fixed layout not updating inside a nested scroll container after its classic scrollbar disappears when using line-clamp with an absolutely positioned descendant. (324368@main) (189451731)
  • Fixed :dir() matching for a <bdi> element without a dir attribute so it updates when its ancestor’s direction changes. (323439@main) (189472957)
  • Fixed light-dark() inside image-set() incorrectly accepting a bare string as an image URL. (323705@main) (189627071)
  • Fixed CSSStyleSheet.replaceSync() on a detached constructed stylesheet appending rules instead of replacing them, which could also corrupt other stylesheets sharing the same contents. (323736@main) (189634428)
  • Fixed CSSMatrixComponent.toMatrix() returning the component’s internal DOMMatrix for 3D matrices, so mutating the result no longer changes the component. (323948@main) (189653766)
  • Fixed a positioned inline-block after the line-clamp clamp point remaining visible. (323402@main) (189672687)
  • Fixed -webkit-min-device-pixel-ratio and -webkit-max-device-pixel-ratio media features serializing as min--webkit-device-pixel-ratio and max--webkit-device-pixel-ratio. (323731@main) (189687289)
  • Fixed a flex item with aspect-ratio and a calc() height resolving to a zero flex base size. (323464@main) (189753886)
  • Fixed CSSStyleDeclaration.setProperty() to override an existing !important custom property when called without a priority. (324083@main) (189765758)
  • Fixed a table flex item with borders being wider than its content and borders. (323471@main) (189766215)
  • Fixed line-clamp: auto to account for collapsed block-end margins when finding the clamp point. (323468@main) (189767053)
  • Fixed a table flex item with borders or padding and a non-auto min-width being too wide. (323466@main) (189768684)
  • Fixed the grid-template shorthand so empty [] line names before a row’s leading line names no longer overwrite the previous row’s track size. (323665@main) (189769011)
  • Fixed line-clamp: auto to find the clamp point before balancing content. (323481@main) (189774214)
  • Fixed calc() negation of a typed-arithmetic product flipping the sign of every factor. (323577@main) (189779826)
  • Fixed object-fit: none and object-fit: scale-down to size an image without natural dimensions against its content box. (323503@main) (189844957)
  • Fixed the content size suggestion of grid items with a preferred aspect ratio to be clamped correctly in grid layout. (323655@main) (189848571)
  • Fixed: Changed the auto value of the continue property to normal to match the specification. (323557@main) (189893145)
  • Fixed the grid layout engine to compute collapsed tracks for repeat(auto-fit). (323658@main) (189937668)
  • Fixed a regression where word-break: keep-all allowed many incorrect line breaks. (324360@main) (189973862)
  • Fixed the placement of outside list markers in vertical writing modes. (323721@main) (190017212)
  • Fixed grid layout to collapse empty repeat(auto-fit) tracks and their gutters. (323843@main) (190062840)
  • Fixed grid layout to use the transferred size suggestion for the automatic minimum size of grid items. (323855@main) (190065310)
  • Fixed the intrinsic minimum block size of a column flex item being incorrectly clamped by a smaller max-block-size. (323872@main) (190077511)
  • Fixed selectors with mixed-case attribute names failing to match HTML elements in XHTML documents when used with ancestor combinators. (323947@main) (190089102)
  • Fixed CSSUnitValue for container query length units reporting a number type instead of a length type. (323949@main) (190095825)
  • Fixed the linear() easing function rejecting a percentage written before the number in a control point. (323951@main) (190097099)
  • Fixed calc() type inversion overflowing its exponent, which could produce incorrect results. (323873@main) (190099592)
  • Fixed CustomStateSet add() and delete() invalidating style when the set does not change. (323982@main) (190183292)
  • Fixed the CSSCounterStyleRule.name setter not updating the name getter or cssText. (324049@main) (190185460)
  • Fixed a grid item with a non-auto min-block-size overflowing the grid when the grid shrinks and the item’s content wraps. (324127@main) (190192125)
  • Fixed quote marks generated by open-quote and close-quote not taking the ::first-line style. (324380@main) (190252758)
  • Fixed a stretched item in a column-reverse wrapping flex container ignoring its main-axis auto margin. (324398@main) (190272880)
  • Fixed calc-size(auto, size) making an aspect-ratio box wider than its max-content size when a child has a percentage height. (324649@main) (190727077)
  • Fixed an absolutely positioned box inside content after the line-clamp clamp point not being laid out. (324457@main) (190732934)
  • Fixed an orthogonal block-level box inside an inline box incorrectly adding its top and bottom margins to its container’s width. (324652@main) (190767962)

Canvas

New Features

  • Added the lang attribute to CanvasTextDrawingStyles. (324596@main) (143391479)
  • Added support for fontKerning, fontStretch, fontVariantCaps, and textRendering in canvas text drawing styles. (324110@main) (190388702)
  • Added support for the inherit value of the direction attribute in CanvasTextDrawingStyles. (324603@main) (190908024)

Resolved Issues

  • Fixed black pixel artifacts when drawing a Canvas 2D image through WebGL after applying a gradient. (324218@main) (189661903)
  • Fixed drawing a VideoFrame in BGRX format to ignore the alpha component. (323809@main) (189794952)

Forms

Resolved Issues

  • Fixed appearance: base-select option hover styling being clipped when rotated transforms are applied. (323903@main) (171933742)
  • Fixed the date picker not using the correct appearance when automatic dark mode is applied. (324046@main) (188747558)
  • Fixed <selectedcontent> elements so they update at the correct times in an enhanced <select>. (324149@main) (189480559)
  • Fixed deselecting the selected option of a dropdown <select> so it always selects the first option. (323929@main) (190021473)
  • Fixed: Updated the default <select> styles to match upstream changes for better interoperability with other browsers. (324587@main) (190724295)

HTML

Resolved Issues

  • Fixed the HTML parser to treat <search> as a special element, so it closes an open <p> element correctly. (323915@main) (190117661)

Deprecations

  • Removed support for the rightmargin and bottommargin presentational attributes on <body>. (323444@main) (189727708)

Images

New Features

  • Added support for object-fit, object-position, and object-view-box on SVG documents in <embed> and <object>. (323822@main) (190118512)

Resolved Issues

  • Fixed the drag image to match the rendered image when dragging an image. (323399@main) (189759601)
  • Fixed the size of images without natural dimensions, such as SVG images, when they are written to the pasteboard. (323529@main) (189873103)
  • Fixed the placement of an SVG image that specifies only a width when used with object-fit: none or object-fit: scale-down. (323567@main) (189909016)
  • Fixed inconsistent sizing of SVG images that have a width or height of 0. (323884@main) (190171855)
  • Fixed an SVG image with a width or height less than 1 failing to load. (324048@main) (190315280)
  • Fixed an issue where 8-bit grayscale and CMYK images were converted to 16-bit images when drawn, wasting memory. (324389@main) (190488500)
  • Fixed image URLs containing a fragment directive so that only the fragment before the directive is used, such as box in foo.svg#box:~:param(--bg, green). (324650@main) (190961757)

JavaScript

Resolved Issues

  • Fixed Math.sumPrecise returning an incorrect result for -Number.MAX_VALUE. (324530@main) (178356895)
  • Fixed a module worker evaluating its entry script a second time when the worker imports its own URL. (323507@main) (188303176)
  • Fixed GetPrototypeFromConstructor to read prototype before calling GetFunctionRealm, matching the specification. (323408@main) (189773842)
  • Fixed an incorrect optimization in the DFG compiler that could produce wrong results for code using the logical NOT operator (!). (323639@main) (189876762)
  • Fixed an incorrect result in optimized code for the integer conversion used by operations such as Math.trunc() and array index handling. (323624@main) (189876905)
  • Fixed an incorrect optimization in the DFG compiler for right shift operations, which could produce wrong results when the shift count exceeds the 32-bit mask. (323595@main) (189886500)
  • Fixed an issue where calling RegExp as a function could be incorrectly optimized into creating a new regular expression object. (323678@main) (189954203)
  • Fixed await a ** b and the use of await as an identifier in a module to correctly throw a SyntaxError. (323662@main) (189958423)
  • Fixed labelled function declarations so they are declared in the enclosing scope. (323675@main) (189972133)
  • Fixed the "use strict" directive to reject strict mode reserved words as parameter and function names. (323677@main) (189973461)
  • Fixed TypedArray and ArrayBuffer constructors to throw a RangeError for a negative length. (323740@main) (189976051)
  • Fixed parsing of let, async of, and destructuring declarations in for loop heads. (324319@main) (190415260)
  • Fixed includes, indexOf, and lastIndexOf on BigInt64Array and BigUint64Array matching BigInt values outside the element type’s range. (324313@main) (190436212)
  • Fixed Array.prototype.push called with no arguments to throw an error when the array’s length is not writable. (324204@main) (190437846)
  • Fixed private class brand installation to happen before class elements are evaluated. (324222@main) (190509577)
  • Fixed module code to report an error when an import or export declaration redeclares a var. (324223@main) (190509652)
  • Fixed Map, Set, WeakMap, and WeakSet constructors to accept undefined or null as the iterable when optimized by the DFG compiler. (324418@main) (190657930)
  • Fixed an issue where private name checks (#x in obj) were compiled incorrectly by the DFG compiler. (324397@main) (190663814)
  • Fixed an issue where an overwritten length property on sloppy-mode arguments objects was not converted using ToLength. (324407@main) (190676161)
  • Fixed Array.prototype.sort calling bound comparator functions incorrectly when sort is inlined. (324512@main) (190788820)

Media

Resolved Issues

  • Fixed the caption font size menu in media controls not changing the caption font size. (324327@main) (137012134)
  • Fixed a high CPU usage issue where media rendering repeatedly busy-looped the main thread while a seek or flush was pending. (323669@main) (188769771)
  • Fixed encrypted media playback enqueuing samples that were blocked on a key and submitted before a flush or seek. (323532@main) (189588721)
  • Fixed text-decoration styles not being applied to the inner background element of WebVTT cues. (323874@main) (190139268)
  • Fixed the spatial video renderer not coming back after media controls are re-enabled. (323986@main) (190170810)
  • Fixed Media Source Extensions playback permanently stalling after a video track is flushed. (324032@main) (190198848)
  • Fixed the video viewer’s native controls incorrectly applying YouTube-specific behavior. (324150@main) (190242515)
  • Fixed previously enabled subtitle tracks not being re-enabled when switching the subtitle menu to off inline and then back on in fullscreen. (324522@main) (190417162)
  • Fixed audio sample timing to respect trimmed timing information. (324387@main) (190519875)
  • Fixed MediaRecorder losing its final data when stopped while a timeslice data request was in flight. (324405@main) (190561436)
  • Fixed WebVTT cue parsing to require exact keyword matches for alignment settings, rejecting values that only begin with a valid keyword. (324444@main) (190662853)
  • Fixed unicode-bidi styles set with ::cue not applying to the WebVTT cue background. (324479@main) (190682540)
  • Fixed WebVTT cue line positioning so the snap-to-lines step is computed from the cue’s first line box, correcting vertical cue rendering. (324447@main) (190685928)
  • Fixed an issue where media time values with equal negative numerators and different timescales were compared in the wrong order. (324514@main) (190698694)
  • Fixed WebCodecs AudioDecoder to decode Opus content with more than two channels. (324615@main) (190709046)
  • Fixed FLAC audio decoding always failing in WebCodecs. (324640@main) (190923178)

Networking

Resolved Issues

  • Fixed WKWebsiteDataStore reusing its connection pool after proxyConfigurations changes, which caused requests to bypass the new proxy. (324143@main) (179411114)
  • Fixed the image loader’s lookup of available images to use the partitioned memory cache. (323594@main) (189778787)
  • Fixed non-ASCII Last-Event-ID request header values being UTF-8 encoded twice for EventSource requests. (323875@main) (189977696)
  • Fixed EventSource to no longer treat a UTF-16 byte order mark as valid. (324395@main) (190634672)
  • Fixed XMLHttpRequest JSON responses to always be decoded as UTF-8. (324483@main) (190717643)

Performance

New Features

  • Improved performance by caching document.querySelector() results. (323620@main) (189789022)
  • Improved navigation performance by starting navigations in a cached Cross-Origin-Opener-Policy process before the response arrives. (323631@main) (189790518)

Resolved Issues

  • Fixed a regression where filtered layers with composited ancestor clips allocated oversized filter buffers. (324517@main) (180473291)
  • Fixed slow repaints and long hangs when registering many small highlights with the CSS Custom Highlight API. (324157@main) (186478678)

Rendering

Resolved Issues

  • Fixed selection gaps not being repainted when clearing a selection in vertical-rl writing mode. (323935@main) (14699394)
  • Fixed font fallback to select fonts and glyphs by canonical equivalence. (323560@main) (165656837)
  • Fixed glyph bounds for complex text to include shaped glyph origins, preventing clipped or mispositioned text. (323777@main) (184235153)
  • Fixed frames that use custom scrollbars continually resizing. (323645@main) (186222322)
  • Fixed composited layer bounds to account for the filter outsets of a layer and its ancestors, preventing filtered content from being clipped. (323864@main) (187924396)
  • Fixed missed overlap for layers inside a stacking-context composited scroller nested in another composited scroller. (323793@main) (189067776)
  • Fixed incorrect inline layout after removing content when partial layout damage started at an inline item of the removed box. (323458@main) (189801792)
  • Fixed composited child elements being misplaced when their compositing ancestor gains or loses a child clipping layer. (323487@main) (189836116)
  • Fixed an issue on macOS where scroll bars in nested iframes kept the old style when the system scroller style changed. (324289@main) (189965521)
  • Fixed uninitialized blurred areas being drawn for blur styles on Cocoa platforms. (324350@main) (190254152)
  • Fixed the text width cache returning incorrect widths when kerned and unkerned versions of a font are used repeatedly. (324093@main) (190360672)
  • Fixed videos being clipped to the wrong corner-shape when an ancestor also has a rectangular clip. (324288@main) (190402270)
  • Fixed an issue where an animated SVG <img> scaled below its natural size left repaint trails. (324343@main) (190485225)
  • Fixed blurred areas being drawn with uninitialized content for some blur styles. (324350@main) (190563103)
  • Fixed the last-line baseline of a block container whose last in-flow child is a scroll container so that the baseline sits at the child’s bottom margin edge, preventing misaligned text. (324653@main) (190772090)

SVG

New Features

  • Added support for referrerPolicy on the SVG <a> element. (323659@main) (135952757)
  • Added support for the SVG 2 text layout algorithm. (324088@main) (190288583)

Resolved Issues

  • Fixed <textPath> not being laid out again when the d attribute changes on a referenced <path> that has display: none. (323938@main) (189579922)
  • Fixed animated percentage values in SVG text y and dy resolving against the viewport width instead of its height. (323847@main) (189739351)
  • Fixed the SVG timeline resetting to 0 when it is paused and seeked before it begins. (324084@main) (189739894)
  • Fixed <use> copies of an SVG <a> keeping the old target after the target attribute changes. (324087@main) (189740877)
  • Fixed SMIL full clock values with more than 1,193,046 hours overflowing during parsing. (323932@main) (189760312)
  • Fixed SVG <image> elements so a referenced SVG image treats its own root preserveAspectRatio as none. (323568@main) (189909967)
  • Fixed SMIL repeat(n) conditions never firing after seeking past repeat iterations. (323927@main) (189986502)
  • Fixed the SVGLength.value setter to convert font-relative units using up-to-date style. (324072@main) (189988135)
  • Fixed an issue where changing the children of an SVG <mask> used by CSS mask-image did not repaint the masked element. (324026@main) (190101845)
  • Fixed SVG <font-face> rules being registered in a non-deterministic order. (324089@main) (190353779)
  • Fixed repaint rect computation for SVG text with vector-effect: non-scaling-stroke so the stroke is repainted correctly per text run. (324095@main) (190363935)
  • Fixed composedPath() in an event listener on a <use> element exposing its user agent shadow root. (324329@main) (190528878)
  • Fixed <feImage> to place and draw its referenced image the same way as an SVG <image> element. (324592@main) (190881899)

Security

New Features

  • Added console logging when Cross-Origin-Opener-Policy or Cross-Origin-Embedder-Policy isolation is expected but not achieved. (324426@main) (190563075)

Resolved Issues

  • Fixed clicking in a cross-origin iframe incorrectly activating same-origin descendants of its ancestors. (323897@main) (181549362)
  • Fixed a regression where creating a <select> element triggered a Content Security Policy style-src-attr violation. (324283@main) (189869066)
  • Fixed frames in an about:blank popup not being treated as secure contexts. (323985@main) (190126073)
  • Fixed data: documents so that only top-level data: documents are treated as non-secure contexts. (324086@main) (190309192)
  • Fixed Geolocation to be blocked in non-secure contexts. (324227@main) (190455029)

Web API

New Features

  • Added support for the estimated line item type in Apple Pay JS. (324470@main) (187363098)
  • Added support for VideoFrame.copyTo() converting to RGB on the fly. (323866@main) (189797745)
  • Added support for the colorSpace option in VideoFrame.copyTo(). (323952@main) (189902908)

Resolved Issues

  • Fixed document.fullscreenEnabled and document.webkitFullscreenEnabled to return consistent values. (323549@main) (102937855)
  • Fixed element.checkVisibility() to work for unslotted elements. (324155@main) (118648432)
  • Fixed the AAC AudioEncoder producing an incorrect decoder config description. (324436@main) (164878480)
  • Fixed getComposedRanges() over-rescoping the range when a passed shadow root is a descendant of the boundary’s shadow tree. (324385@main) (180810983)
  • Fixed the Navigation API so navigation.back() and navigation.traverseTo() on the main frame land on the correct session history entry after a subframe has navigated. (324589@main) (187959384)
  • Fixed element.checkVisibility() returning false for <option> and <optgroup> elements in a list box. (323990@main) (189211521)
  • Fixed getBoundingClientRect() returning incorrect boxes for <option> and <optgroup> elements in a list box. (324424@main) (189227828)
  • Fixed VideoFrame, AudioData, EncodedVideoChunk, EncodedAudioChunk, MediaStreamTrack, and MediaStreamTrackHandle so they can no longer be cloned or transferred into a different agent cluster. (323558@main) (189747796)
  • Fixed VideoFrame.copyTo() to succeed for BGRX and RGBX frames. (323569@main) (189797144)
  • Fixed Media Session artwork images that have no intrinsic size, such as SVG without width, height, or viewBox, so they are sized using the default sizing algorithm with a 512×512 ideal artwork size. (323461@main) (189822449)
  • Fixed VideoFrame.copyTo() to allow converting to an RGB format when VideoFrame.format is null. (323953@main) (189932876)
  • Fixed an issue where a page navigated to about:blank lost its history entry when the next navigation occurred. (324566@main) (190237322)
  • Fixed an issue where a previous tap could cause a later tap to be treated as a hover. (324259@main) (190357260)
  • Fixed Blob‘s text() to no longer sniff UTF-16 byte order marks when decoding. (324134@main) (190382159)
  • Fixed a navigation to the current URL reporting the wrong navigationType when document.domain is set. (324269@main) (190420617)
  • Fixed Blob‘s text() to decode its data as it arrives. (324356@main) (190449988)
  • Fixed location.replace() on a frame in another process so it no longer adds a session history entry. (324264@main) (190512610)
  • Fixed an issue where setInterval() could be scheduled using a stale cached next fire time. (324432@main) (190658646)
  • Fixed an issue where navigating an iframe away from its initial about:blank document by clicking a link added a history entry. (324562@main) (190687356)
  • Fixed requestIdleCallback() idle period scheduling by using the actual frame period at 60Hz, so idle periods are no longer scheduled right before the frame deadline. (324545@main) (190768899)

Deprecations

  • Deprecated XML external entities, and added a console warning when they are used. (324200@main) (184870654)

Web Extensions

Resolved Issues

  • Fixed tabs.onUpdated to not report unchanged values and to report rapid property toggles. (324357@main) (84204907)
  • Fixed Safari showing no error when the declarativeNetRequest rule_resources JSON is not a list. (323963@main) (146113788)
  • Fixed calling action.setPopup() with an invalid popup path leaving the popup unrecoverably broken. (324142@main) (166048992)
  • Fixed non-repeating browser.alarms alarms not being removed from the alarm map after firing. (323750@main) (173399426)
  • Fixed action.setIcon() throwing an error when tabId is null instead of setting the global icon. (324277@main) (176975795)
  • Fixed runtime.openOptionsPage() so it focuses the existing options page when it is already open. (324541@main) (177344808)
  • Fixed browser.alarms.clearAll() to return a Promise<undefined>. (323534@main) (187115327)
  • Fixed alarms.onAlarm to report the next scheduledTime for repeating alarms when the background page was unloaded. (323752@main) (189951123)
  • Fixed an issue where an extension’s own content rule lists were applied to pages other than its extension pages. (323863@main) (190036270)
  • Fixed content blocker rules that combine request-method with a top URL or frame URL condition not working. (324050@main) (190316684)

Web Inspector

New Features

  • Added display of ::slotted() styles in the Styles sidebar of the Elements tab. (323654@main) (29035069)
  • Added support for exposing Shadow DOM parts in the Elements tab. (323685@main) (100250231)
  • Added reporting of rendered font character ranges. (324359@main) (143740107)
  • Added visual distinction for private methods in the Console. (324399@main) (190664079)

Resolved Issues

  • Fixed ::placeholder styles for input elements not showing up in the Styles sidebar. (323670@main) (56514581)
  • Fixed a regression where console.group and console.groupCollapsed did not always encapsulate grouped messages. (323613@main) (142028751)
  • Fixed toString being called twice when logging an object with the console. (323668@main) (169883641)
  • Fixed a discrepancy where Web Inspector and WebDriver reported different computed roles for the same element. (323965@main) (179841782)
  • Fixed: Improved Web Inspector performance by no longer iterating all rendering frame records to calculate the overview graph height. (323614@main) (181921116)
  • Fixed the Elements tab to collect CSS variables for both styles and keyframes. (323598@main) (189934225)
  • Fixed the Elements tab to correctly parse animation names when editing keyframes. (323600@main) (189934494)
  • Fixed console screenshots appearing in read-only Timelines recordings. (323599@main) (189934498)
  • Fixed markers appearing in read-only Timelines recordings. (323601@main) (189934611)
  • Fixed the Timelines tab to stop updating the time when capture is inactive. (323604@main) (189936051)
  • Fixed Web Inspector to handle URLs with an empty host. (323603@main) (189936372)
  • Fixed Web Inspector to parse URLs with ports and blob: URLs in stack traces. (323606@main) (189936384)
  • Fixed the Timelines tab to show capture progress only for the active recording. (323607@main) (189936469)
  • Fixed the Elements tab to correctly parse animation keys when editing keyframes. (323608@main) (189936580)
  • Fixed the Console’s await convenience being included in code completion requests. (323605@main) (189936602)
  • Fixed the Timelines tab to preserve imported heap snapshots when exporting recordings. (323609@main) (189936631)
  • Fixed stack trace popovers to reposition when the Web Inspector window is resized. (323611@main) (189936790)
  • Fixed Web Inspector to distinguish the source code of multiple Worker instances. (323612@main) (189936821)
  • Fixed awaited evaluation in the Console to preserve the selected call frame scope. (323610@main) (189937067)
  • Fixed the Elements tab to refresh styles after a CSS grouping rule edit fails. (323616@main) (189937131)
  • Fixed the Elements tab to also track inherited CSS variable dependencies. (323623@main) (189939830)
  • Fixed Web Inspector to report errors that occur while loading source maps. (323653@main) (189956520)
  • Fixed the Elements tab overlay for SVG <text> nodes appearing in the wrong position. (324055@main) (190158402)
  • Fixed Web Inspector so that edits to a style sheet mutate the live style sheet contents. (323979@main) (190235686)
  • Fixed links to inline CSS in Web Inspector incorrectly adding the starting column of the <style> element. (323998@main) (190248080)
  • Fixed Web Inspector so that the client nodes of a -webkit-canvas are always an array. (323999@main) (190248443)
  • Fixed Web Inspector so that a script is associated with its resource even when the MIME type is incorrect. (324003@main) (190249110)
  • Fixed Web Inspector canvas recording so that explicit null arguments are recorded. (324002@main) (190249115)
  • Fixed a Worker created after disabling All Requests breakpoints still pausing in Web Inspector. (324004@main) (190249260)
  • Fixed the debugger to use the sourceURL for breakpoints when one is provided. (324169@main) (190465797)
  • Fixed checkbox navigation items being shifted down. (324172@main) (190465799)
  • Fixed frames missing from the Timelines tab. (324171@main) (190465833)
  • Fixed Call Trees in the Timelines tab sometimes missing content. (324173@main) (190466008)
  • Fixed focus being lost in the Timelines tab when using arrow keys to select a data grid row that is out of view. (324174@main) (190466428)
  • Fixed focus being lost in the Graphics tab when using arrow keys to select a tree element that is out of view. (324175@main) (190466777)
  • Fixed Safari extension scripts missing from the Sources tab in Web Inspector. (324509@main) (190768385)

WebAssembly

Resolved Issues

  • Fixed one-type recursion groups to match their shorthand form. (323455@main) (189807549)
  • Fixed atomic loads and stores in the BBQ and OMG compilation tiers to be compiled consistently with the interpreter. (324185@main) (190473201)
  • Fixed any.convert_extern to accept a subtype of externref. (324243@main) (190498804)

WebGL

Resolved Issues

  • Fixed: Improved WebGL memory use by allocating default framebuffer storage when it is first used instead of at context creation. (323911@main) (188901577)
  • Fixed WebGL not working in shared workers and service workers in some conditions. (324220@main) (189900485)
  • Fixed: Improved WebGL performance by reusing drawing buffers once the compositor releases them. (323882@main) (190042989)
  • Fixed resizing a canvas that has not been drawn to unnecessarily allocating the default framebuffer. (324144@main) (190308555)
  • Fixed single-sample rendering that uses derivatives incorrectly forcing the sample mask output on Metal. (324221@main) (190417440)
  • Fixed an issue where a GPU reset caused all subsequently created WebGL contexts to stop working. (324346@main) (190510529)

WebGPU

New Features

  • Added support for the texture-component-swizzle feature, allowing createView() to take a swizzle option that permutes or fills texture channels. (324543@main) (189135744)

Resolved Issues

  • Fixed GPU memory for a destroyed GPUTexture not being released while a GPUBindGroup that referenced it still exists. (323838@main) (189420456)
  • Fixed: Improved performance of reading back an offscreen canvas that uses WebGPU. (323834@main) (189530498)
  • Fixed WGSL uniformity analysis to treat textureDimensions() of a write-only or read-write storage texture as non-uniform. (324225@main) (190384655)
  • Fixed copying ImageBitmap content to a texture in WebGPU. (324352@main) (190516486)

WebRTC

Resolved Issues

  • Fixed an issue where a camera or microphone permission denial was remembered even when it did not come from the user. (324309@main) (190551948)
  • Fixed camera and microphone permission requests to be queued so they are presented one at a time. (324473@main) (190618582)

November 05, 2026 12:09 AM

November 03, 2026

Introducing Adaptive Dark Mode

Surfin’ Safari

We’re excited to announce Adaptive Dark Mode in Safari 27.3 beta. When enabled by a user, Safari adapts light websites and presents them as dark.

In the spring of 2019, Safari 12.1 became the first browser to support dark mode on the web. A lot of people love using dark mode, but not every website has a dark version. Almost 40% of web page loads are from users in dark mode, and yet only 10% of page loads support it. If you love dark mode, you’ve probably hit a site that felt particularly bright. Adaptive Dark Mode is designed to close that gap, providing a consistent dark experience across the web.

A demo website in light mode on the left and dark mode on the right. It's a complex site with lots of UI full of bits of color that's transformed by Adaptive Dark Mode.

You can try it today in Safari 27.3 beta. To enable, first make sure your system appearance is set to dark. Then browse to a website with a bright background. Safari will offer a preview of Adaptive Dark Mode. Tap ‘Try’.

For websites that signal support of dark mode with <meta name="color-scheme" content="light dark"> in HTML or color-scheme: light dark in CSS, Adaptive Dark Mode does not change a thing. For the rest of the web, it quickly analyzes the page, looking at colors and estimating whether it’s predominantly light or dark. If the page generally has a dark background with light text, nothing is changed. When a page has a light appearance, Adaptive Dark Mode adapts the page to be dark.

How does Adaptive Dark Mode work?

At the core of Adaptive Dark Mode is the transformation of color. Sometimes colors are simply decorative, while other times they have semantic meaning. Adaptive Dark Mode determines when to invert colors so elements remain visible or when to leave colors unchanged.

While creating Adaptive Dark Mode, we developed a color matrix to invert the luminosity of colors while preserving hue. It’s used to make light backgrounds dark, and dark text light. When transforming text color, Adaptive Dark Mode aims for sufficient contrast as defined by the WCAG 4.5:1 contrast ratio. Light CSS gradients are also transformed to dark.

Adaptive Dark Mode makes careful judgments so transformed content looks good and stays usable: buttons have contrast against backgrounds, text is visible over images, and hover menus look clear.

It takes the same care with images. Most are left unchanged, preserving their integrity. But many icons and logos can lose contrast when the background color behind them is inverted. So Adaptive Dark Mode inspects SVG images and transforms colors when needed. For bitmap images, it analyzes the pixel histograms and inverts simple black-and-white logos.

Side by side screenshots of the Interop 2026 dashboard in Safari. On the left, the original light mode website. On the right, the site transformed by Adaptive Dark Mode. The graph still matches light mode (untransformed) but everything else on the page is dark.

Effects like masks and mix-blend-mode can create undesirable results on an element whose colors have been transformed, so in some cases Adaptive Dark Mode turns such effects off. Also, <canvas> content is left unchanged, since canvases contain a wide range of content that’s hard to predict.

It’s precisely because Adaptive Dark Mode is happening deep within the WebKit browser engine that we are able to find creative ways to solve the most complex challenges, apply the transformations with great speed and power efficiency, all while rigorously protecting user privacy.

Making your site work even better

In general, a web page will work well with Adaptive Dark Mode if it follows good modern web development practices.

If your site still uses older techniques, small updates can help. For logos and icons, swap bitmap images for SVG. Use solid or gradient colors in CSS instead of tiled background images. Replace image sprites with modern CSS for rounded corners, shadows, and button states. Create content images with transparent backgrounds instead of baking the page background color into the image. Add padding to containers to prevent crowding.

Perhaps you want a specific dark mode experience for a certain portion of the page. Apply color-scheme: light dark to the relevant element and add your own dark styles.

Building your own dark mode

For complete control over how your site looks in dark mode, add full support. It’s a terrific time to do so. It’s never been easier.

The demo website from above, only now, with a third browser window showing what that website might look like with a custom dark mode.
Here’s an example of a website in light mode (left) and with Adaptive Dark Mode applied (center). If the site creates a custom dark mode, it could look like the third version (right).

All the web technology needed to fully support dark mode is widely available today, including:

Newer CSS functions like color-mix() and contrast-color(), plus popular techniques for organizing design systems with CSS variables provide additional options and flexibility.

Be sure to declare dark mode support using the color-scheme meta tag in your HTML head and/or by setting the color-scheme property on :root or body in your CSS. Don’t wait to inject this information later via JavaScript. This best practice helps ensure your site is rendered correctly as fast as possible.

Using the “light only” color scheme

If you would like your web app or website to not be transformed by Adaptive Dark Mode, you can communicate this using color-scheme. The HTML <meta name="color-scheme" content="light only"> tag lets browsers know that 1) this page supports light mode and 2) should only be displayed in light mode. The color-scheme: light only rule does the same thing, but with the flexibility of CSS, allowing you to just target specific elements on the page — or of course, you can target the whole page with :root { }. Support for the ‘only’ keyword arrived in Safari 27.3 beta 2.

Protecting user privacy

Apple firmly believes user privacy is a fundamental human right. If websites could detect when a user has enabled Adaptive Dark Mode, that would increase the amount of fingerprintable information and weaken privacy protections against trackers.

As a result, Safari prevents websites from detecting when Adaptive Dark Mode has been applied. JavaScript that queries CSS colors will get the original authored colors, not the transformed ones. Any images drawn into canvas render in their original form.

Feedback

If you manage a website or web app, and would like to discuss how Adaptive Dark Mode is adapting your project, feel free to reach out at web-evangelist@apple.com.

If you try out Adaptive Dark Mode as a user, and find a site that does not work well, we’d love for you to file a report with Feedback Assistant. (Learn more about filing Feedback).

November 03, 2026 06:00 PM

November 02, 2026

Igalia WebKit Team: WebKit Igalia Periodical #84

Igalia WebKit

Update on what happened in WebKit in the week from October 26 to November 2.

In this week's installment we have an improvement to WebKit when running under Flatpak, OpenSSL being made the default crypto backend for the WPE and GTK ports, improved support for the Media Playback Quality spec, and the usual plethora of improvements and fixes to the Layer-Based SVG Engine.

Cross-Port 🐱

When running in Flatpak, use the flatpak-portal D-Bus API to spawn sub processes instead of flatpak-spawn. This provides a small improvement in launch times and memory usage.

OpenSSL is now the default backend for crypto in the WPE and GTK ports. The libgcrypt backend will remain in tree for the current cycle but can be considered deprecated and will be removed after the next stable release series.

Multimedia 🎥

GStreamer-based multimedia support for WebKit, including (but not limited to) playback, capture, WebAudio, WebCodecs, and WebRTC.

Improved support for the Media Playback Quality spec in WebKitGTK/WPE, avoiding statistics counter resets during flushes and pauses, and properly keeping the statistics after player suspension.

Graphics 🖼️

Fixed the sizing of <foreignObject> in vertical writing modes in the Layer-Based SVG Engine (LBSE), where its width and height were swapped. The viewport size is now mapped to the logical axes, as the legacy SVG engine does.

Skipped SVG filters that change nothing, such as an feOffset without an offset or an feDropShadow with zero blur and offset, in the Layer-Based SVG Engine (LBSE), matching the legacy SVG engine.

Fixed missing repaints in the Layer-Based SVG Engine (LBSE) when an attribute of an SVG filter primitive changes before the filter has painted for the first time. The elements using the filter are now always repainted, as in the legacy SVG engine.

Removed an extra clip on the outermost <svg> element in the Layer-Based SVG Engine (LBSE), which truncated drop-shadow() filters, box-shadow, outlines painting outside the border box, and rotated inline <svg> elements. The existing viewport overflow clip already handles this correctly, including transforms.

Fixed getScreenCTM() in the Layer-Based SVG Engine (LBSE) to include CSS zoom on the outermost <svg> while still ignoring page zoom, matching the legacy SVG engine. This resolves the remaining getScreenCTM() issues in LBSE.

Removed the transform-to-root check from SVG container layout in the Layer-Based SVG Engine (LBSE), which re-laid out a whole subtree whenever an ancestor changed its transform to the outermost <svg> element. Transforms now invalidate the relevant state directly, so these extra re-layouts are no longer needed.

Fixed outlines, masks and event regions being displaced on an outermost <svg> that is not at the origin of its containing block in the Layer-Based SVG Engine (LBSE). Besides the visual fix, this also corrects the regions used for wheel event handling.

Fixed SVG text being laid out with a stale font size after its own transform changes in the Layer-Based SVG Engine (LBSE). The transform is now updated before the on-screen font size is computed, so text metrics match the legacy SVG engine.

Fixed non-scaling strokes on SVG text ensuring they are measured in the coordinate system of the outermost <svg> element, like shapes already were, instead of in screen space. A CSS transform on the outermost <svg> or on one of its HTML ancestors now scales the vector-effect: non-scaling-stroke of text correctly, in both the Layer-Based SVG Engine (LBSE) and the legacy SVG engine.

Aligned the filter source image to whole device pixels when repainting filtered SVG content in the Layer-Based SVG Engine (LBSE), so a repaint always fully replaces the pixels it touches, instead of leaving stale pixels behind when content wasn't device-pixel aligned.

That’s all for this week!

By Igalia WebKit Team at November 02, 2026 07:30 PM

October 26, 2026

Igalia WebKit Team: WebKit Igalia Periodical #83

Igalia WebKit

Update on what happened in WebKit in the week from October 19 to October 26.

An exciting bag of news this week which includes a fix for processes not being closed under memory pressure, process prewarming when sandboxing is enabled, a couple of GStreamer-related changes including one that make auxiliary processes' startups faster, and the usual influx of LBSE fixes. In addition to that we got a libsoup release with security fixes, and the deprecation of the legacy libwpe-based API.

Cross-Port 🐱

Fixed prewarmed and suspended processes not being closed under memory pressure.

Fixed process prewarming when the sandbox is enabled.

Networking 📶

Networking support, including the libsoup HTTP library.

Released libsoup 3.8.1. The release contains various security fixes as well as bug fixes for HTTP/2 and zstd encoding support.

Multimedia 🎥

GStreamer-based multimedia support for WebKit, including (but not limited to) playback, capture, WebAudio, WebCodecs, and WebRTC.

The Media player now internally uses the GStreamer playbin3 element, playbin support is deprecated given that it isn't much maintained anymore and less efficient than playbin3.

Dynamically load flite when used, reducing process startup time significantly.

Graphics 🖼️

Fixed SVG text not being re-measured when a CSS transform changes the scale of an enclosing <g>, <use> or nested <svg> in the Layer-Based SVG Engine (LBSE). Such text used to keep its old on-screen font size, and it is now laid out again at the new scale, as it already was when the SVG transform attribute changed.

WPE WebKit 📟

The legacy libwpe-based API has been deprecated. While removal is not planned, this means that starting with the 2.56 release series (due in March 2027) building applications that are use it will produce deprecation warnings. The recently announced WPEPlatform API should be preferred, and migrating to it is encouraged.

WPE Platform API 🧩

New, modern platform API that supersedes usage of libwpe and WPE backends.

The formats array passed to the WPEClipboard.changed virtual function now indicates the array length correctly. While at it, the virtual functions of the WPEClipboard class have been documented.

Community & Events 🤝

Fix dark-mode when using wpe-platform-gtk.

Infrastructure 🏗️

Build with -z pack-relative-relocs reducing the binary size and memory usage.

That’s all for this week!

By Igalia WebKit Team at October 26, 2026 07:03 PM

WebKit Features for Safari 27.2

Surfin’ Safari

Safari 27.2 is here, refining Customizable Select, renaming a Grid Lanes property, and delivering 20 bug fixes across accessibility, media, CSS, and more.

Customizable Select

Customizable Select originally landed in Safari 27.0. Since then, the web standards process continues to polish details. And implementations across all browsers are being updated to match.

For decades, when you use the <select> element and a user selects an option, it gets displayed in the “button” (the interactive part of the control that appears on the page). The new superpowers of Customizable Select include a <selectedcontent> element to give you more control over what’s displayed and how it looks.

<select id="status">
  <button>
    <selectedcontent></selectedcontent>
  </button>
  <option>Processing</option>
  <option>Shipped</option>
  <option>Delivered</option>
  <option>Canceled</option>
</select>

But you are not required to use <selectedcontent>. You can still write HTML like this:

<select id="status">
  <option>Processing</option>
  <option>Shipped</option>
  <option>Delivered</option>
  <option>Canceled</option>
</select>

Now in Safari 27.2, the new ::field-content pseudo-element gives you the ability to style the text of the selected option that’s displayed on the page, without needing the <selectedcontent> element.

In addition, Safari 27.2 refines a few of the default UA styles for appearance: base-select. The button’s contents vertically center more reliably. The selected option text now inherits text-overflow from <select>. The way we draw the ::picker-icon has been updated so it can have a more straightforward layout. And the button uses flex: 1 to manage spacing, instead of margin-inline-start: auto on ::picker-icon, so the picker icon is always visible, even without <selectedcontent>.

Also, previously a large select picker could render partly outside the viewport. Safari 27.2 keeps it in view by applying safe alignment. This comes from a broader change to CSS alignment: for absolutely and fixed positioned boxes, safe now keeps the box within its original containing block. You can also now combine the safe and unsafe keywords with normal alignment.

CSS

Grid Lanes originally shipped in Safari 26.4 last March. Since then, the CSS Working Group decided to rename the flow-tolerance property to fit-tolerance to better describe its purpose. Safari 27.2 updates Grid Lanes to use the new name. Meanwhile, the older name will keep working for now as a deprecated alias. Be sure to update your code soon. Safari Web Inspector can help by marking what needs to be changed.

What is the fit-tolerance property? It controls how picky the browser is when it chooses where to put the next item. Without any tolerance, the browser looks for the column (or row in the other direction) whose content precisely ends highest on the page, and puts the item there. But if several columns end at nearly the same height, you may not care which one wins. With fit-tolerance, you define how much of a difference counts. When columns are within that distance of each other, the browser treats them as tied and just picks the first one. The default is fit-tolerance: 1em.

A higher tolerance keeps consecutive items in neighboring columns, instead of jumping back and forth across the page to chase tiny differences in height. That makes for a better experience tabbing through content, especially for sighted keyboard users, who watch focus move from item to item. So most of the time, you should increase this number significantly. How much depends on your content sizes and how much they vary.

Play with fit-tolerance on the Field Guide to Grid Lanes to better understand what it does. Learn more about how Grid Lanes and Fit Tolerance work in our article, Introducing CSS Grid Lanes.

Bug fixes and more

In addition to these updates, in our continuing commitment to quality, WebKit for Safari 27.2 includes 20 bug fixes:

Accessibility

  • Fixed list, table, tree, and canvas content being omitted from the accessible name or description when the containing element is referenced by aria-labelledby or aria-describedby. (186015032)
  • Fixed VoiceOver repeating text it had already announced when a page streams content by re-rendering an aria-live region instead of appending to it. (186766347)
  • Fixed an issue where VoiceOver did not identify <input type="date">, <input type="time">, and <input type="color"> controls as pop-up buttons on iOS. (187788949)

Animations

  • Fixed animation-range-start and animation-range-end resolving incorrectly when zoom is applied, including when the value is inherited from an element with a different zoom factor. (185760120)

CSS

  • Fixed an issue where an anchor-positioned element inside a modal <dialog> could be laid out offscreen and become invisible when the page was scrolled. (187933878)

Forms

  • Fixed: Switched default content for the::picker-icon in Customizable Select to render with a glyph that has a more straightforward layout, no longer requiring negative margins to align correctly. (187411081)

HTML

  • Fixed the window load event never being dispatched when a readystatechange handler starts a new load. (186475302)

JavaScript

  • Fixed for await...of over an async generator skipping the Promise.prototype.constructor lookup required by PromiseResolve when that property is replaced. (183960626)

Media

  • Fixed an issue where AudioContext.decodeAudioData() failed with an EncodingError on valid AAC/M4A audio files. (183474728)
  • Fixed video showing a black screen while audio continued to play after the video decoder was invalidated. (184041554)
  • Fixed ManagedMediaSource not firing the startstreaming event when seeking to a time that is not buffered, which could stall playback. (185166787)
  • Fixed an issue where a media element’s play() promise was not resolved when a SourceBuffer contained clear, unencrypted content. (188022510)
  • Fixed an issue where the WebCodecs VideoEncoder could stall and produce no output for H.264 Main and High profiles. (188201298)

Networking

  • Fixed cookies set by a top-level navigation’s own response being capped to a seven-day lifetime when the site’s hostname is CNAMEd to a different registrable domain. (183787115)

SVG

  • Fixed dynamic changes to a <foreignObject> element’s x and y values having no effect. (184273832)

Safari MCP

  • Fixed Safari taking keyboard focus from other apps when creating tabs or capturing screenshots under Safari MCP control. (182558329)

Web API

  • Fixed an issue where transient activation could be incorrectly discarded, causing APIs gated on user activation (such as the Clipboard API) to fail on the first attempt. (187872193)

Web Extensions

  • Fixed a significant delay before web extension content is injected when Safari is cold launched. (185268822)

WebGPU

  • Fixed a severe frame rate drop when calling executeBundles() with render bundles that contain drawIndirect() or drawIndexedIndirect(). (186099290)

WebRTC

  • Fixed an issue where getUserMedia() requests for audio permanently failed with NotAllowedError after the system audio service was reset. (186939802)

Updating to Safari 27.2

Safari 27.2 is available on iOS 27.2, iPadOS 27.2, visionOS 27.2, macOS 27.2 Golden Gate, macOS Tahoe and macOS Sequoia. On iOS, iPadOS, visionOS, and macOS 27.2 Golden Gate, you can update to Safari 27.2 as part of the OS update in Settings > General > Software Update. On older versions of macOS, if you want to only update Safari, you can do so in Software Update, by clicking “more”.

Feedback

We love hearing from you. To share your thoughts, find our web evangelists online: Jen Simmons on Bluesky / Mastodon, Saron Yitbarek on Bluesky / Mastodon, and Jon Davis on Bluesky / Mastodon. You can follow WebKit on LinkedIn. If you run into any issues, we welcome your feedback on Safari UI (learn more about filing Feedback), or your WebKit bug report about web technologies or Web Inspector. If you run into a website that isn’t working as expected, please file a report at webcompat.com. Filing issues really does make a difference.

October 26, 2026 06:30 PM

October 23, 2026

WebKit Features for Safari 27.1

Surfin’ Safari

iPhone Duo, looking amazing, partially folded like an open book. Photo of a woman on the left, app icons on the right, including Safari in the Dock.

Today iPhone Duo arrives, and with it, Safari 27.1.

The “big news” here is no news. It’s Safari, with the same WebKit engine and the same web technologies as every other iPhone. Your websites and web apps work on iPhone Duo just like they do everywhere else.

Safari 27.1 does bring one change to users. Now, when someone holds iPhone Duo partially folded like a book while looking at a PDF on the web, Safari 27.1 automatically shows PDFs as two pages side by side. When they open the screen flat or close it fully, Safari switches back to one page.

For web developers, Safari 27.1 on iPhone Duo includes a preview of the Device Posture API and the Viewport Segments API, available behind feature flags.

Viewport sizes

As a web designer or developer, the first question you might have is about the sizes & shapes of the screens.

Little drawings of iPhone Duo in different poses. Outer display. Tent pose. Inner display landscape. Book pose. Inner display portrait. Laptop pose.

If you have been making websites and web apps using modern responsive web design techniques to support a wide range of different sized screens with media queries and flexible layouts using CSS Grid, Flexbox & more, your project is going to look and work great!

For those wondering, here’s more detail about viewport sizes in various poses… iPhone Duo has two displays: the outer display when the phone is closed, and the larger inner display when the phone is open. The outer display works in portrait or landscape, with the tab bar at the bottom by default, just like other iPhones. The inner display also works in both orientations, with the tab bar shown at the top, similar to iPad or Mac. When the inner display is in landscape, users can also put two windows side by side, which gives each one a narrow portrait viewport. And just like iOS on other iPhones, when the user scrolls, the tab bar shrinks in size to put the focus on the web content.

Across the possible combinations, the viewport width (in CSS pixels) in Safari 27.1 can range from 382 to 951. And the height can range from 368 to 902.

Width SVH LVH
Outer display, portrait 382 580 618
Outer display, landscape 594 368 406
Inner display, portrait (with one tab) 669 867 902
Inner display, landscape (with one tab) 951 585 620
Inner display, each window with two side by side 385 571 609

However, you don’t need to worry about all these numbers. Instead, lean into the wisdom of Responsive Web Design techniques. Create layouts for your websites and web apps that simply work on every sized screen from the extremely large to the very tiny. Modern CSS has powerful tools that make flexible, fluid, responsive layouts incredibly easy. Avoid UA sniffing to detect devices, and use feature detection instead. These approaches put your project in the best position to just work anywhere, no matter what the future brings.

You don’t need iPhone Duo in hand to test your website or web app. Download Xcode 27.1 or later to get Device Hub, where you can create and use a simulator for iPhone Duo.

Creating webpage layouts for a folding screen

When iPhone Duo is partially folded, it opens up some interesting design possibilities. Some web developers have been curious about the Device Posture API and the Viewport Segments API. Both are supported in Safari 27.1 on iPhone Duo as a developer preview behind feature flags.

To enable, go to Settings > Apps > Safari > Advanced > Feature Flags, and turn on both “Device Posture API” and “Viewport Segments API”.

If you try out these features, let us know what you think. Do they unlock the ability to create the kind of experiences you want to create for iPhone Duo? Would you want them to work a bit differently instead?

Device Posture API

The Device Posture API tells you whether or not the user’s screen is partially folded — in both CSS and JavaScript with the device-posture media query and navigator.devicePosture. Both report one of two postures:

  • folded, when the device is partially folded, in a book or laptop pose
  • continuous, when the screen is flat — which includes both fully open and fully closed displays

In CSS, use the media query to change the styles when the device is partially folded:

@media (device-posture: folded) {
  .layout {
    display: grid;
    grid-template-columns: 1fr 1fr;
  }
}

Or in JavaScript, use navigator.devicePosture.type to read the current posture and listen for the change event to know when it changes.

function reportPosture() {
  console.log(`Device posture: ${navigator.devicePosture.type}`);
}

reportPosture();
navigator.devicePosture.addEventListener("change", reportPosture);

Viewport Segments API

The Viewport Segments API is more complex. It describes the “segments” of the display — the separate portions on either side of the fold.

Once you turn on the feature flag, Safari 27.1 supports the horizontal-viewport-segments and vertical-viewport-segments media queries. They match the number of segments laid out side by side, or stacked top to bottom:

@media (horizontal-viewport-segments: 2) {
  /* Code that only applies when the device is held like a book, 
     with two side-by-side segments */
}

@media (vertical-viewport-segments: 2) {
  /* Code that only applies when the device is held like a laptop, 
     with two vertically stacked segments */
}

To help you size things to the exact edges of each segment, the Viewport Segments API adds six environment variables:

  • viewport-segment-width
  • viewport-segment-height
  • viewport-segment-top
  • viewport-segment-right
  • viewport-segment-bottom
  • viewport-segment-left

Each takes two indexes: the segment’s column, then its row, counting from 0 at the top left. When two segments sit side by side, the left one is 0 0 and the right one is 1 0. When they’re stacked, the top one is 0 0 and the bottom one is 0 1.

For example, this layout creates two columns, each the width of the available space on each side of the fold:

@media (horizontal-viewport-segments: 2) {
  .wrapper {
    display: grid;
    grid-template-columns:
      env(viewport-segment-width 0 0)
      env(viewport-segment-width 1 0);
  }
}

In JavaScript, window.viewport.segments gives you an array of DOMRect objects, one per segment. Check it again whenever the posture changes or the window resizes:

function reportSegments() {
  window.viewport.segments.forEach((segment, i) => {
    console.log(`Segment ${i}: ${segment.width}px × ${segment.height}px`);
  });
}

reportSegments();
navigator.devicePosture.addEventListener("change", reportSegments);
window.addEventListener("resize", reportSegments);

With the Device Posture API and the Viewport Segments API behind flags, there may still be time to evolve how they work. Let us know if you have thoughts or requests.

Web content under the controls

You might have noticed the outer display of iPhone Duo is wider and shorter than other iPhone displays. To give content more vertical space, iOS moves controls to the side.

An example website on iPhone Duo, the outer (smaller) display in portrait. The webpage fills most of the screen, except for a column on the right. That space holds the controls for Safari and for the phone itself. It fills with the same light purple color of the website.

In Safari 27.1, web content does not currently extend to fill the space behind these controls. Just like on other platforms, Safari looks at the web page and extends its background color to fill the space.

We would like to give web designers and developers the ability to design and use that space in more creative ways. Perhaps you want to extend your header background across the full width of the display, for example. It might seem like there’s just one obvious way to make this possible, but really there are many possibilities.

We could use the HTML viewport meta tag with viewport-fit="cover" to extend the webpage into that space, hoping developers also use the appropriate environment variables to scoot important content away from system controls. But as we research how today’s sites are built, we see a lot of projects that fail to inset all four edges correctly. We are investigating if there’s a better way to accomplish the same result — something more foolproof and perhaps easier for you to use. If you have ideas, we’d love to hear them. And stay tuned.

What do you think?

To share your thoughts with us, reach out to Jen Simmons on Bluesky / Mastodon. If you run into any issues, we welcome your feedback about Safari UI (learn more about filing Feedback), or your WebKit bug report for web technologies or Web Inspector.

October 23, 2026 05:00 PM

October 22, 2026

Release Notes for Safari Technology Preview 255

Surfin’ Safari

Safari Technology Preview Release 255 is now available for download for macOS Golden Gate and macOS Tahoe. If you already have Safari Technology Preview installed, you can update it in System Settings under General → Software Update.

This release includes WebKit changes between: 322011@main…323397@main.

Accessibility

Resolved Issues

  • Fixed VoiceOver presses dispatching a click event with an event.detail of 0. (322978@main) (187732224)
  • Fixed radio buttons in a group to be exposed as required when the required attribute is set on one of the buttons. (322894@main) (189100525)
  • Fixed list box accessibility children not updating when options are added or removed. (323117@main) (189412211)

Animations

Resolved Issues

  • Fixed getComputedStyle() ignoring the pseudo-element when checking whether a property is being animated. (322681@main) (188502400)
  • Fixed requestAnimationFrame() handles to use the full unsigned long range defined by the IDL. (322922@main) (189270148)
  • Fixed view transition setup to flush the update callback queue. (323173@main) (189533643)

CSS

New Features

  • Added support for CSS module scripts, allowing a stylesheet to be imported from JavaScript using import sheet from "./styles.css" with { type: "css" }. (322887@main) (80917329)

Resolved Issues

  • Fixed font-family: initial acting like font-family: inherit instead of resolving to the initial value. (322304@main) (54306281)
  • Fixed box-sizing: content-box being ignored for percentage-height tables (display: table), which caused vertical padding to be dropped. (322812@main) (122587242)
  • Fixed min-width: max-content being ignored for the min-content contribution of a percentage-width replaced element. (322974@main) (179947068)
  • Fixed the top color extension to use relative colors with filters when applicable. (323153@main) (186098400)
  • Fixed Range.getBoundingClientRect() returning empty rects for content inside a content-visibility: hidden subtree. (323204@main) (186165755)
  • Fixed elements using a background image that depends on currentColor not repainting when the current color changes. (323240@main) (188036420)
  • Fixed subgrid implicit grid-area line names being matched by the first substring instead of the suffix. (322238@main) (188203288)
  • Fixed tree-scoped var() and custom function references inside an @function failing to resolve when the function is called from a :host rule. (322679@main) (188206030)
  • Fixed an issue where duplicate anchor-name values among <model> elements in a spatial portal resolved unpredictably. (322709@main) (188209540)
  • Fixed @scope so that structural DOM changes correctly invalidate scope membership when the scoping root selector depends on them. (322680@main) (188398757)
  • Fixed revert inside @starting-style dropping @starting-style declarations from lower cascade layers. (322688@main) (188463947)
  • Fixed the rotate property to preserve the sign of the axis when serializing. (322682@main) (188493811)
  • Fixed revert and revert-layer using the wrong cascade origin for visited and unvisited link styles, which wrongly disabled native appearance. (323205@main) (188494505)
  • Fixed getComputedStyle() serializing the overscroll-behavior shorthand as a single value of the larger axis instead of both axis values. (322685@main) (188534035)
  • Fixed block-axis multicolumn content being painted at the wrong offset when the column set does not start at the top of the multicol container. (322687@main) (189067503)
  • Fixed shape-margin values above 65535px on an image shape-outside producing a much smaller float area. (322727@main) (189069223)
  • Fixed a line-clamp container to establish an independent formatting context. (322767@main) (189090409)
  • Fixed replaced elements ignoring aspect-ratio when both width and height are auto. (322745@main) (189096923)
  • Fixed the line-clamp block ellipsis being placed after a <wbr> in a white-space: nowrap line. (322758@main) (189103913)
  • Fixed an overflowing preserved tab not wrapping to the next line. (322759@main) (189104880)
  • Fixed the line-clamp block ellipsis not hanging with hanging-punctuation. (322772@main) (189105782)
  • Fixed text-overflow truncating a unicode-bidi: plaintext line in the wrong direction. (322866@main) (189108555)
  • Fixed an image shape-outside dropping an above-threshold run that starts in the last column. (322756@main) (189116534)
  • Fixed line-clamp: auto not being able to clamp between sibling blocks. (322836@main) (189124610)
  • Fixed line-clamp: auto clamping inside a box with a set height. (322842@main) (189138321)
  • Fixed the line-clamp block ellipsis being placed on a line that only contains a float. (322791@main) (189151667)
  • Fixed the baseline of an empty MathML <mpadded> element to account for border-top and padding-top. (322826@main) (189209081)
  • Fixed CSS grid intrinsic width computation to resolve auto-fill repetitions against the grid container’s minimum and maximum size. (322848@main) (189223527)
  • Fixed CSS grid column sizing to resolve percentage rows against the block size rather than the inline size when estimating row sizes. (322850@main) (189230010)
  • Fixed grid container minimum and maximum sizes to respect box-sizing: border-box. (322856@main) (189235347)
  • Fixed a percentage-width child of a vertical writing mode box to keep following the box’s width after a style change. (322964@main) (189316242)
  • Fixed cross-fade() to match the CSS Images Level 4 specification and the behavior of other browsers. (322918@main) (189356185)
  • Fixed grid layout to return the grid’s used block size, so the container’s size is correct once layout is complete. (323053@main) (189357098)
  • Fixed grid containers to resolve percentage and calc() values for min-width, max-width, min-height and max-height. (323051@main) (189358198)
  • Fixed :visited style not updating for links inside shadow trees. (323396@main) (189380898)
  • Fixed baseline alignment of grid items so it is applied after the items have been sized. (323238@main) (189457596)
  • Fixed attr() to avoid an unnecessary legacy code path in simple cases. (323098@main) (189485680)
  • Fixed MediaList writes being silently dropped after the owning rule or stylesheet is destroyed. (323203@main) (189496021)
  • Fixed collapsed SVG shapes and images being painted when visibility: collapse should be treated as hidden. (323110@main) (189496450)
  • Fixed block boxes after the line-clamp clamp point remaining visible. (323220@main) (189543433)
  • Fixed content before the line-clamp clamp point losing its lines when it does not change. (323225@main) (189569400)
  • Fixed block boxes after the line-clamp clamp point still taking up space. (323253@main) (189570948)
  • Fixed floats after the line-clamp clamp point remaining visible. (323337@main) (189578173)
  • Fixed adoptedStyleSheets continuing to apply removed sheets after its length is shrunk, and ignoring changes to a sheet after its duplicate is removed. (323233@main) (189579103)
  • Fixed an implicit @scope rule in a constructed stylesheet adopted by a document matching nothing. (323237@main) (189584948)
  • Fixed floats making the line-clamp container taller. (323349@main) (189630233)
  • Fixed floats adding to the scrollable overflow of the line-clamp container. (323353@main) (189641420)
  • Fixed line-clamp: auto containers having no line clamp until the clamp point is found. (323357@main) (189652879)
  • Fixed floats painting past the block-end content edge of the line-clamp container. (323390@main) (189653161)
  • Fixed container size queries truncating fractional container sizes to integers. (323328@main) (189653185)
  • Fixed absolutely positioned elements whose containing block is an inline after the line-clamp clamp point remaining visible. (323395@main) (189671905)

Editing

Resolved Issues

  • Fixed DataTransfer treating any format that merely starts with text/plain, such as text/plainfoo, as text/plain. (322954@main) (189265567)

Forms

New Features

  • Added support for <button> in the native appearance of <select>. (322716@main) (178876621)

Resolved Issues

  • Fixed :open no longer matching a <select> whose base appearance picker is still showing. (322666@main) (188911581)
  • Fixed a <select> element computing its selected option’s style from outdated ancestor styles. (322665@main) (189109118)
  • Fixed a vertical range slider fill overshooting its value when tick marks are present and the thumb is hidden. (322780@main) (189116567)

HTML

Resolved Issues

  • Fixed a <style> element’s sheet to be built from its child text content instead of descendant text. (322128@main) (188047912)
  • Fixed tabbing to a popover invoker to move focus to the popover itself when the popover is focusable. (323172@main) (188267440)
  • Fixed module <script> elements with an empty or invalid src to fire the error event asynchronously instead of synchronously. (323198@main) (189384155)
  • Fixed setting src to the empty string on a connected <script> element to fire an error event. (323200@main) (189385352)
  • Fixed <script> elements with an ASCII-whitespace-only src being treated as having an empty src. (323236@main) (189386063)

JavaScript

Resolved Issues

  • Fixed Object.defineProperty to read descriptor fields after converting the property key, so side effects from key conversion are respected. (322982@main) (179389067)
  • Fixed an issue where errors thrown by another realm’s built-ins such as Number, String, and Math functions came from the caller’s realm after DFG optimization. (323152@main) (189057417)
  • Fixed catch parameters named await to be rejected as a syntax error in contexts where await is not an identifier. (322702@main) (189183089)
  • Fixed await to be treated as an identifier in a non-async function nested in the parameters of an async function. (322703@main) (189183268)
  • Fixed unary + so it no longer changes the evaluation order of operands in binary expressions. (322817@main) (189267513)
  • Fixed constant array literal elements to be evaluated only once when emitting bytecode for array literals. (322943@main) (189295197)
  • Fixed instanceof with a bound function reading the target’s prototype property when the left-hand side is a primitive. (322956@main) (189388195)
  • Fixed Array.fromAsync to get next from the async iterator before constructing the this value. (322967@main) (189402953)
  • Fixed using and await using declarations being rejected inside a function nested in a switch case or default clause. (323061@main) (189472811)
  • Fixed for-in skipping indexed properties of functions. (323086@main) (189504414)

Media

Resolved Issues

  • Fixed remote audio participants no longer autoplaying on web sites using real-time calls. (323320@main) (182922583)
  • Fixed WebVTT cue display so the backdrop no longer overflows outside of the cue. (323325@main) (183472002)
  • Fixed an issue where video using Media Source Extensions never started playing on some sites, such as disneyplus.com. (322877@main) (184182838)
  • Fixed WebCodecs ignoring the profile, bit depth, and chroma subsampling fields in VP9 codec strings. (322962@main) (188868426)
  • Fixed the WebCodecs AV1 encoder reporting support for 10-bit, 12-bit, monochrome, and 4:4:4 or 4:2:2 codec strings that it always encodes as 8-bit 4:2:0. (322975@main) (188882755)
  • Fixed WebCodecs computing incorrect plane layouts for I422 and I444 VideoFrame objects. (322972@main) (188885351)

Networking

Resolved Issues

  • Fixed failed modulepreload resource loads getting stuck in the cache by retrying them. (323305@main) (124280450)
  • Fixed a regression where XMLHttpRequest and fetch sent an empty request body for a disk-backed File when the page is controlled by a service worker. (323318@main) (183071407)

Rendering

Resolved Issues

  • Fixed a block-level box with aspect-ratio inside a float growing when its content changes. (322742@main) (188872195)
  • Fixed selection gaps not being painted for a line whose selection top is above its block container. (322750@main) (189113277)
  • Fixed an issue where an element’s box size was incorrect when a padding change was followed by another style change before layout. (322929@main) (189319402)
  • Fixed an issue where changing a table cell’s width did not resize its row in a table with a vertical writing mode. (322934@main) (189322207)

SVG

New Features

  • Added support for the side attribute on the SVG <textPath> element. (323115@main) (174710714)
  • Enabled the inline layout engine for SVG text. (323141@main) (189532515)

Resolved Issues

  • Fixed zero intrinsic size not being respected for SVG in <object> and <embed> elements. (322529@main) (132152736)
  • Fixed SVG elements inside hidden containers being focusable. (322650@main) (171590890)
  • Fixed an issue where changing a list attribute detached the SVGNumber, SVGLength, and SVGPoint objects held by a script instead of updating them. (323046@main) (186835842)
  • Fixed changing only paint-order not triggering a repaint of SVG content. (322013@main) (188533466)
  • Fixed SVG animations continuing to run when the animation element or an ancestor fails conditional processing. (322829@main) (188780260)
  • Fixed svgView() fragments with an invalid or zero-size viewBox drawing the SVG unscaled. (323217@main) (189611839)

Security

Resolved Issues

  • Fixed external scripts blocked by the strict-dynamic Content Security Policy keyword to fire an error event. (323191@main) (189386852)

Web API

New Features

  • Added support for the cross-origin-isolated Permissions Policy feature. (322663@main) (189124490)
  • Added support for transferring streams to an AudioWorklet. (323327@main) (189677199)

Known Issues

  • document.activeElement returns the <html> element instead of null for HTML documents with no body, which differs from other browsers and breaks the WebDriver Get Active Element command. (323385@main) (190343977)

Resolved Issues

  • Fixed an issue where interactive-widget=resizes-content did not reflect the correct innerHeight, causing the sign-in screen on kidshibbett.com to disappear from view. (186817825)
  • Fixed pointermove firing on the parent of a pointerover target that is removed during pointerover. (322451@main) (188882772)
  • Fixed Cross-Origin-Embedder-Policy not being ignored in non-secure contexts. (322838@main) (189193030)
  • Fixed Range.deleteContents() and Range.extractContents() removing nodes outside the range when the range end is moved during the operation. (322920@main) (189262302)
  • Fixed the element.dataset named getter to no longer match non-data-* attributes or invalid property names. (322936@main) (189263319)
  • Fixed NodeIterator leaving its reference on a detached node when an ancestor of the reference node with no following node in the root is removed. (322933@main) (189263890)
  • Fixed requestIdleCallback() handles above 2^31 so they can be cancelled with cancelIdleCallback(). (322942@main) (189266546)
  • Fixed secure context status being determined after document creation instead of when the document is created. (322939@main) (189268799)
  • Fixed the document’s default event handler running twice for bubbling events targeted at the document, which fired DOMActivate twice. (323065@main) (189379133)
  • Fixed requestIdleCallback callbacks no longer running after removing a same-origin iframe that had a pending setTimeout. (323148@main) (189379748)
  • Fixed Window-only and DedicatedWorkerGlobalScope-only types being deserialized in other kinds of workers. (323120@main) (189423491)
  • Fixed moveBefore() leaving selectors that depend on the child list, such as :has() and :empty, stale after a move. (323044@main) (189461850)
  • Fixed FileSystemHandle objects being deserialized in non-secure contexts. (323219@main) (189588333)
  • Fixed SpeechRecognitionResultList being exposed in non-secure contexts. (323326@main) (189631080)
  • Fixed DOMException and QuotaExceededError not being exposed in every global scope. (323313@main) (189677643)

Deprecations

  • Removed the unimplemented document.implementation.createCSSStyleSheet() method. (322805@main) (142045904)

Web Extensions

New Features

  • Added support for browser.notifications.update() in web extensions. (322071@main) (187615870)
  • Added support for browser.notifications.clear(), browser.notifications.getAll(), and browser.notifications.getPermissionLevel() in web extensions. (322297@main) (187616262)

Resolved Issues

  • Fixed storage.local.get() returning the same array or object instance for distinct empty arrays and objects in a stored value. (323003@main) (189204323)
  • Fixed extension page subframes to have unpartitioned storage access. (323034@main) (189301936)

Web Inspector

New Features

  • Added a CSS keyframe animation editor to Web Inspector. (322820@main) (15801500)
  • Added stack trace popovers to Web Inspector. (322748@main) (20582649)
  • Added evaluation previews to the Web Inspector console. (322754@main) (182773473)
  • Added basic WGSL syntax highlighting. (322649@main) (189147091)
  • Added Web Inspector Graphics instrumentation for updates to the label of WebGPU devices and pipelines. (322793@main) (189247758)

Resolved Issues

  • Fixed Find in the Elements tab not matching HTML elements and attributes when the search text contains a space character. (323277@main) (109289803)
  • Fixed Log Message breakpoint actions to log objects as rich console arguments instead of plain text. (322733@main) (116907593)
  • Fixed the Infos and Debugs scope bar items disappearing from the console after the console is cleared. (323269@main) (125621103)
  • Fixed await not being supported in the console in some scenarios. (323272@main) (127381329)
  • Fixed Styles to correctly account for declarations marked !important within a CSS @layer rule when computing style overrides. (323275@main) (151931708)
  • Fixed console log messages dismissing the DOM Event popover. (323306@main) (153842239)
  • Fixed Beacon resources in the Sources tab showing an error instead of the response content. (323270@main) (170805067)
  • Fixed the wrong source code location being shown for inline scripts with a source map. (322747@main) (182487329)
  • Fixed Web Inspector to compare resolved resource types after receiving a response. (323274@main) (183988946)
  • Fixed Web Inspector to no longer automatically invoke native getters on functions and prototype objects. (323271@main) (184191697)
  • Fixed the Web Inspector frame selector popup disappearing when paused in the debugger. (322739@main) (186908754)
  • Fixed Web Inspector to clear the active call frame when its target is removed. (322734@main) (186984661)
  • Fixed Web Inspector not offering JavaScript runtime completions for BigInt values. (322731@main) (186985019)
  • Fixed Web Inspector to preserve explicit execution context selections across back-forward navigation. (323281@main) (186985083)
  • Fixed Web Inspector to preserve newer execution context selections during navigation. (323282@main) (186985387)
  • Fixed the Network emulation tab bar icon appearing inverted in dark mode in Web Inspector. (322729@main) (187758172)
  • Fixed the Styles panel showing var() with a non-ASCII custom property name, such as var(--色), as invalid even though it resolves. (322076@main) (188169850)
  • Fixed the debugger leaving no active call frame when the active target resumes while another target is still paused. (322843@main) (188827760)
  • Fixed Web Inspector to only update the currently active Timeline recording. (322738@main) (189207884)
  • Fixed the Elements tab to include the host element when pushing a pseudo element. (322871@main) (189321868)
  • Fixed Web Inspector to complete rendering notifications before the next rendering update. (322873@main) (189322496)
  • Fixed back/forward cache restore incorrectly reporting a navigation failure after a pagehide beacon. (322983@main) (189350210)
  • Fixed Web Inspector reporting “domain already enabled” errors when going back to a page restored from the back/forward cache after a process swap. (323239@main) (189375630)
  • Fixed the loaderId changing when a navigation continues in another process. (323167@main) (189567072)
  • Fixed the selected node not being highlighted when hovering keyframe headers in the Elements tab. (323260@main) (189669693)
  • Fixed the Elements tab offering Duplicate Selector, Add Rule, and Create Rule context menu actions for keyframes. (323263@main) (189669755)
  • Fixed filtering styles in the Elements tab incorrectly hiding matching keyframe sections. (323262@main) (189669803)
  • Fixed the Network tab reported response timing not accounting for network throttling. (323261@main) (189669852)
  • Fixed the Storage tab hiding or mishandling entries with an empty key or value. (323267@main) (189669876)
  • Fixed importing a Timeline recording modifying more than the imported recording. (323268@main) (189670026)
  • Fixed search results missing matches on consecutive lines. (323264@main) (189670067)
  • Fixed revealing a source location from the Console selecting the wrong location in a pretty printed script. (323265@main) (189670127)
  • Fixed the Console missing known parameters for some native DOM methods. (323307@main) (189689587)
  • Fixed an uncaught ReferenceError for row in the Layout Timeline overview graph. (323308@main) (189689709)
  • Fixed an uncaught TypeError in the Console when formatting table parameters for a logged message with no parameters. (323363@main) (189724207)

WebGL

Resolved Issues

  • Fixed WebGL default framebuffer queries to be answered at the WebGLRenderingContext level. (322700@main) (188899947)
  • Fixed an issue where resizing a canvas with separate width and height assignments could lose the WebGL context. (323298@main) (188900757)
  • Fixed a WebGL canvas becoming blank after its drawing buffer was resized when painted without accelerated compositing on Windows. (322870@main) (189321624)
  • Fixed full clears of renderbuffers being treated as partial clears under robust resource initialization. (323300@main) (189506198)
  • Fixed renderbuffers being initialized on the CPU in the Metal backend. (323297@main) (189606864)

WebRTC

Resolved Issues

  • Fixed RTCCertificate objects being deserialized in workers. (323182@main) (189499857)

October 22, 2026 09:22 PM

October 21, 2026

Nikolas Zimmermann: Making LBSE performant

Igalia WebKit

Closing the MotionMark/Suits gap between LBSE and the legacy SVG engine

October 21, 2026 12:00 AM

October 19, 2026

Igalia WebKit Team: WebKit Igalia Periodical #82

Igalia WebKit

Update on what happened in WebKit in the week from October 12 to October 19.

Another busy week for the team! This week we had a variety of fixes, including the regular batch on the Layer-Based SVG Engine, a new pair of properties for WPE WebKit, and a new ApplicationManifest API which improves support for progressive web applications.

Cross-Port 🐱

Add new ApplicationManifest API. This allows looking up the webmanifest for Progressive Web Applications.

Graphics 🖼️

Fixed the repaint rects of SVG text using vector-effect: non-scaling-stroke, first by taking the scale into account in the Layer-Based SVG Engine (LBSE), which used to repaint a larger area than it paints for scaled-up text. A follow-up made the decision per text run, since vector-effect is not inherited and a <tspan> inside a non-scaling <text> paints a regular scaling stroke, and unified the computation so that the legacy SVG engine and LBSE now report identical repaint rects.

Fixed the composited bounds of layers below pixel-moving CSS filter effects such as blur() and drop-shadow(), whose backing store previously missed the pixels the filter samples outside the clip. The clips above a layer are now expanded by the filter outsets of the layer and its ancestors, and a change to those outsets updates the composited bounds of all descendants. This change fixes the cross-platform part, and the CoreAnimation based ports, as well as GTK/WPE.

Fixed repaints when an SVG element switches to a wider clip-path, mask or filter in the Layer-Based SVG Engine (LBSE), which previously only repainted the area it covered before the change and left the newly revealed area stale. Style changes that alter the visual overflow now trigger a repaint of the new bounds as well.

Fixed the repaint region of elements clipped by a <clipPath> with a transform attribute in the Layer-Based SVG Engine (LBSE), where changing that transform repainted the wrong area.

Fixed mix-blend-mode on a document-root <svg> element in the Layer-Based SVG Engine (LBSE), which wrongly blended the page background along with the SVG contents. The blend mode is now applied to the SVG contents only, matching the legacy SVG engine.

Fixed elements staying clipped on screen after their <clipPath> was removed in the Layer-Based SVG Engine (LBSE), because the clipped elements were never repainted. Destroying an SVG resource now notifies its clients, which repaint the area the resource used to cover.

Fixed hit testing of children inside transformed layered SVG groups in the Layer-Based SVG Engine (LBSE), where points outside a shape could hit it and transformed children could be missed entirely.

Stopped recomputing the transform twice on nearly every layout of an SVG container without a layer in the Layer-Based SVG Engine (LBSE). The transform reference box is now remembered for such renderers when transform-box is view-box (the default), so the post-layout update only runs when the box actually moved.

Fixed repaints of SVG groups and their content in the Layer-Based SVG Engine (LBSE), which previously left stale content on screen when something inside a <g> moved or when the <g> itself changed (a new transform / changed clip-path etc.). Images, text and shapes now compute their final position during their own layout, and groups with a layer remember their transform so they repaint both the old and new area on change, which brings text repaints in transformed groups in line with the legacy SVG engine.

Fixed painting of descendants with visibility: visible inside hidden <text> and <foreignObject> elements in the Layer-Based SVG Engine (LBSE), which skipped the whole subtree, unlike the legacy SVG engine. Hidden <g> and <text> containers now also run the mask and clipping mask phases, so such descendants are masked and clipped as well.

WPE WebKit 📟

On WPE expose WindowProperties:geometry and WindowProperties:has-custom-geometry like GTK.

That’s all for this week!

By Igalia WebKit Team at October 19, 2026 07:23 PM

Blocking Identity Resolution Infrastructure

Surfin’ Safari

Apple and WebKit have shipped protections against web tracking for a long time. We put privacy protections in place for cookies in the very first version of Safari, we invented storage and cache partitioning and shipped it ten years before other major browsers, and we’re actively fighting device fingerprinting.

We do this because users deserve to not be tracked, not across websites and not covertly through fingerprinting, unless they explicitly opt in to it. See WebKit’s Tracking Prevention Policy for a comprehensive overview.

Identity Resolution, or the Post-Cookie Era

We fully blocked cross-site cookies six years ago. That was a huge step forward for privacy on the web.

However, the tracking industry has not sat idle since.

A “post-cookie” tracking method emerged, called identity resolution. In this scheme, third parties (i.e., not the website owner) invisibly collect your email address when you log in or sign up for a newsletter. They send the email address to their servers, create an opaque tracking ID permanently linked to your email address, and send that tracking ID back to the webpage. The tracking ID is stored on the website and can from thereon be used for arbitrary purposes linked to your email address, such as building a profile of your browsing or targeting ads at you across websites. Logging out or unsubscribing from that newsletter does not remove the tracking ID or your email address stored with the third party.

Diagram showing four websites (Website 1, 2, 3, and 4) each sending Jane's email address (jane@mail.example) to a central 'Identity resolution server' and receiving back a different tracking ID for each site (Tracking ID 1 through 4), illustrating how an identity resolution service links a shared identifier across multiple sites.

Over time, these third parties collect even more data as you log in using your email address or fill in forms with personal data. They use this to create a data profile of you, known as an identity graph. Here’s an illustrative example of what such a graph may look like:

  Illustration of a person labeled

Email addresses almost always constitute Personally Identifiable Information, or PII. Another example is your phone number. Such information not only allows for cross-site tracking, but also cross-device tracking. For the majority of users, their email address serves as a globally unique ID that they keep for years, even decades. When users provide information for one purpose, it breaks their expectation to use it for another unstated purpose. For example, a user who logs in or signs up for a newsletter is not expecting to have their information sent to a third party for other, potentially long-term, intents.

First party personalization is not tracking, and is not what our technologies seek to prevent. But even if identity graphs are only scoped to data from a single company, i.e. a first party, the central identity resolver is a third party and it holds cross-site, cross-device, and cross-party data about the user and can perform matching and targeting functions in the capacity of a clearinghouse. That is a tracking capability that each first party does not have on their own.

Diagram showing an online store asking 'Which of my users should I promote ice skates to?' A third-party identity resolver in the middle, matching across multiple first-party datasets, contains profiles for Emre, Aline, Mike, and two separate 'Jane' profiles — one from the online store and one from an ice hockey website — which are linked together in a dotted circle labeled 'Match.'

What We Are Doing To Protect Users Against Identity Resolution

Some companies serve as third-party infrastructure to facilitate identity resolution. They are present but invisible on many webpages where you enter your email address or use your email address for login purposes. They are not themselves login providers or newsletter services, they’re just there to track you.

In our iOS, iPadOS, macOS, and visionOS 27 releases, Safari and WebKit block network access to identity resolution infrastructure to prevent cross-site tracking. This stops them from sending your email address to their servers and all the downstream effects of that. The blocking is done on the domain level, meaning they can’t just change the structure of the network request to work around the protection.

The same four-website and identity-resolution-server diagram as before, but each website's request is blocked by a red shield-with-X icon, illustrating Safari blocking the requests and identity resolution being prevented.

Hide My Email

As an iCloud+ subscriber, you are offered the service Hide My Email which enables you to submit unique, randomized email addresses to every new website while still getting their messages in your inbox. Unique, random email addresses provide additional protection against email-based tracking.

Our Commitment to Web Privacy

It’s not enough to block cross-site cookies to prevent cross-site tracking today because the tracking industry has not stopped trying to track users on the web. They have ushered into what they call the post cookie era and in our latest releases we have put strong protection in place. We will continue to prevent covert and cross-site tracking by default, while maintaining the ability for websites to provide first-party personalization to their users.

October 19, 2026 05:50 PM

October 12, 2026

Igalia WebKit Team: WebKit Igalia Periodical #81

Igalia WebKit

Update on what happened in WebKit in the week from October 5 to October 12.

A CSS invalidation bugfix, moving steps for optgroup and hr elements, and a handful of fixes to the Layer-Based SVG Engine compose this week's bag of updates, changing to a calmer pace compared to the packed weeks we've had recently!

Cross-Port 🐱

Fixed various CSS selector invalidation when an element is moved.

Add moving steps for optgroup and hr elements.

Graphics 🖼️

Fixed buffered-rendering="static" for SVG <image> elements in the Layer-Based SVG Engine (LBSE), where the image used to be drawn at the wrong position inside its offscreen buffer and without its own transform. The image is now painted into the buffer relative to its own position, and the buffer is drawn back with the current transformation, which matches the behaviour of the legacy SVG engine.

Added support for pointer-events: bounding-box on SVG text in the Layer-Based SVG Engine (LBSE), so clicks that miss every glyph now fall back to hitting the text's bounding box (see pointer-events). It also fixed hit testing of transformed children in SVG containers, which were being tested a second time without their transforms.

Fixed text not following its path in the Layer-Based SVG Engine (LBSE) when a <textPath> points to a path that doesn't exist yet. The text is now laid out along the path as soon as that path appears in the document, which also covers removing and re-adding the path or changing href to another path.

That’s all for this week!

By Igalia WebKit Team at October 12, 2026 07:21 PM

October 08, 2026

Release Notes for Safari Technology Preview 254

Surfin’ Safari

Safari Technology Preview Release 254 is now available for download for macOS Golden Gate and macOS Tahoe. If you already have Safari Technology Preview installed, you can update it in System Settings under General → Software Update.

This release includes WebKit changes between: 321068@main…322634@main.

Accessibility

Resolved Issues

  • Fixed VoiceOver on iOS not identifying date, time, and color inputs as pop-up buttons. (321384@main) (187696500)
  • Fixed VoiceOver being unable to press buttons, links, and form controls styled with pointer-events: none. (322472@main) (188003870)
  • Fixed VoiceOver announcing every focused ARIA menuitem, menuitemcheckbox, and menuitemradio as “selected”. (322397@main) (188185386)
  • Fixed an issue on iOS where an error message made of several pieces of text was read in reverse order with some text repeated. (322318@main) (188764297)

Animations

Resolved Issues

  • Fixed view timeline animations freezing once a sticky-positioned inline ancestor sticks to the scroll container. (321094@main) (187300111)
  • Fixed an issue where accelerated animations of opacity to 0, including scroll-driven animations, could leave elements fully visible instead of transparent. (322355@main) (187682187)

CSS

New Features

  • Added support for the size, min-size, and max-size shorthand properties. (321710@main) (187495014)
  • Added support for the border-*-radius side shorthands, such as border-top-radius, which set the radii of both corners on one side of a box at once. (322537@main) (187605885)
  • Added support for the interpolate-size property, which allows animations and transitions to and from intrinsic sizing keywords such as auto, min-content, and fit-content. (321370@main) (187701518)
  • Added support for the symbols() function for defining anonymous counter styles. (321624@main) (188048446)
  • Added support for CSS linked parameters, including the link-parameters property and param(), for passing values into linked SVG images. (321822@main) (188217737)
  • Added support for the calc-size() function, which allows calculations on intrinsic sizes such as auto, fit-content, and max-content. (322574@main) (188996389)

Resolved Issues

  • Fixed CSS Grid relayout cost growing quadratically with the number of grid items that have distinct order values. (321853@main) (173962596)
  • Fixed CSS Grid track sizing for items that span multiple tracks to follow the current specification for distributing extra space, including to non-affected tracks. (322402@main) (182725333)
  • Fixed table cells after an empty width: 0 column being placed off-screen. (322376@main) (183452074)
  • Fixed the rendering of outlines with a negative outline-offset on elements that use corner-shape. (321259@main) (186820849)
  • Fixed line-clamp and -webkit-line-clamp to share the same underlying values and cascade correctly against each other, with -webkit-line-clamp now treated as a shorthand. (322292@main) (187100061)
  • Fixed sibling-index() and sibling-count() returning 1 for children of a shadow root. (321182@main) (187201634)
  • Fixed line-clamp, max-lines, and block-ellipsis to match the latest specification syntax, allowing max-lines and line-clamp to combine auto with an integer and block-ellipsis to accept no-ellipsis, ellipsis, or a string. (321832@main) (187371224)
  • Fixed grid track sizing for items that span multiple tracks so that they are sized in groups by span count, as the CSS Grid specification requires. (321194@main) (187413690)
  • Fixed a min-content inline size to give both the min-content and max-content contributions of an element that size. (321278@main) (187601903)
  • Fixed an issue where a stretched grid item whose padding exceeded its grid area was sized narrower than its own padding. (321295@main) (187612019)
  • Fixed implicit named grid lines created by grid-template-areas sometimes being empty. (321401@main) (187753954)
  • Fixed fit-content() grid tracks growing beyond their limit when distributing extra space. (321720@main) (187757242)
  • Fixed grid layout to use a grid item’s margin box, rather than its border box, for its minimum size contribution so items with margins size their tracks correctly. (321614@main) (187850851)
  • Fixed clip-path with a <geometry-box> value ignoring corner-shape and clipping to rounded corners instead. (321557@main) (187891292)
  • Fixed a flex item’s base size ignoring the space available to it when its main size is fit-content. (322557@main) (187932371)
  • Fixed a box with a fixed width ignoring max-width: max-content when contributing its intrinsic size. (322298@main) (188008779)
  • Fixed grid track sizing so a track’s growth limit is never initialized smaller than its base size. (321616@main) (188023571)
  • Fixed the intrinsic width of grid containers with contain: inline-size, container-type: inline-size, or content-visibility: hidden so it comes from contain-intrinsic-size rather than the grid items. (321612@main) (188029152)
  • Fixed missing repaints in CSS Grid layouts when grid items move. (321737@main) (188142676)
  • Fixed table-layout: fixed being ignored when a table’s width is intrinsic, min-intrinsic, or a calc-size() value. (322409@main) (188201473)
  • Fixed relative color syntax to preserve none channel values that are referenced directly, so oklab(from oklab(none none none) l a b) resolves to oklab(none none none). (321724@main) (188215354)
  • Fixed color serialization so that sRGB colors with none components serialize using color(srgb ...), and hsl() and hwb() colors serialize their saturation, lightness, whiteness, and blackness components as percentages. (321758@main) (188254602)
  • Fixed relative color syntax so that missing (none) components are preserved instead of becoming 0 when the origin color is converted to a different color space. (321802@main) (188300348)
  • Fixed relative alpha() colors to serialize in the most precise form using color() syntax, matching color-mix() and other relative color expressions. (321829@main) (188323511)
  • Fixed relative alpha() color functions to require an alpha parameter, so values like alpha(from red) are now invalid, following a CSS Working Group resolution. (321852@main) (188339902)
  • Fixed deprecated CSS system colors, such as ActiveBorder, Menu, Scrollbar, ThreeDFace, and Window, to compute to their CSS Color 4 equivalents on macOS. (322193@main) (188365953)
  • Fixed border-image-repeat: round to round the tile count to the nearest whole number instead of rounding it down. (321996@main) (188369219)
  • Fixed a regression where elements with aspect-ratio inside a column flex container did not shrink when the container got narrower. (321951@main) (188393652)
  • Fixed line-clamp to make room for the block ellipsis during line breaking, moving content that doesn’t fit next to the ellipsis to the next line, including nowrap content. (322495@main) (188393991)
  • Fixed line-clamp so that when the block ellipsis displaces all content on the clamped line, only the ellipsis remains and the line is not made taller by inline boxes. (322500@main) (188440163)
  • Fixed an issue where grid items with aspect-ratio and a percentage height did not shrink when the grid got narrower. (322176@main) (188444964)
  • Fixed CSS color conversion to follow the latest specification rules for missing hue components on achromatic colors, including in hsl() and hwb(). (321984@main) (188471496)
  • Fixed line-clamp adding a block ellipsis to the last line when all of the content fits within the line limit. (322510@main) (188500712)
  • Fixed line-clamp to count the lines of nested blocks towards the maximum number of lines. (322521@main) (188535368)
  • Fixed an issue where a justified line came up short when its hanging trailing whitespace was inside an inline box. (322224@main) (188548682)
  • Fixed line-clamp to remove hanging trailing whitespace before the block ellipsis. (322531@main) (188551752)
  • Fixed an issue where line-clamp was ignored on floats and absolutely positioned boxes. (322551@main) (188552622)
  • Fixed an issue where ruby annotations lost their text inside a line-clamp container. (322578@main) (188552871)
  • Fixed an issue where text inside a display: contents element inherited the wrong style after its parent’s style changed. (322092@main) (188591349)
  • Fixed line-clamp incorrectly clamping lines on a multi-column container. (322584@main) (188595965)
  • Fixed an issue where changing the width of a line-clamp container showed the lines past the clamp point. (322600@main) (188596942)
  • Fixed aspect-ratio being ignored on absolutely positioned boxes with height: fit-content, min-content, or max-content. (322178@main) (188597811)
  • Fixed line-clamp: auto to clamp at the container’s block size. (322603@main) (188621599)
  • Fixed an issue where changing shape-image-threshold did not update the shape when the shape image was a gradient. (322232@main) (188772999)
  • Fixed grid items ignoring their ::first-line styles in certain cases. (322319@main) (188830233)
  • Fixed fit-content() grid tracks with a percentage argument being stretched when the grid’s size is indefinite. (322421@main) (188848301)
  • Fixed overflow-wrap: anywhere keeping extra characters on a line, or breaking inside a surrogate pair, next to punctuation that is prohibited at the start of a line. (322554@main) (188884467)
  • Fixed the initial value of position-visibility to be anchor-visible. (322565@main) (188939499)

Canvas

Resolved Issues

  • Fixed rect() using the wrong winding direction when given a negative width or height. (322101@main) (188036092)
  • Fixed TextMetrics emHeightAscent and emHeightDescent to add up to the font’s em size, and updated the top, hanging, middle, ideographic, and bottom baselines to match when drawing text. (322077@main) (188368945)
  • Fixed sizing of image sources, such as SVG images without a natural size, used by canvas drawImage() and createPattern(), createImageBitmap(), WebGL, and VideoFrame to follow CSS object size negotiation. (322003@main) (188524860)

Deprecations

  • Removed support for the non-standard darker value of globalCompositeOperation. (321811@main) (188035440)

Editing

Resolved Issues

  • Fixed an issue where the caret was hard to see after pasting wrapped text whose color contrasts poorly with the editor’s background, such as dark text pasted into a dark-mode editor. (321615@main) (186543233)
  • Fixed drag-selecting text that starts inside a position: fixed element so that the page autoscrolls when the pointer moves past the edge of the window. (322310@main) (188330363)
  • Fixed an issue where converting a bulleted list to a numbered list left a stray type attribute on the resulting <ol>. (322170@main) (188445470)

Forms

Resolved Issues

  • Fixed ::picker-icon on <select> elements to draw the dropdown arrow as a built-in glyph instead of a text character positioned with a negative margin, so authors no longer need to reset margins when styling it. (321110@main) (185952359)
  • Fixed setting valueAsNumber to throw a TypeError for infinite values and to clear the value when set to NaN, instead of throwing a NotSupportedError for both. (321668@main) (188047421)
  • Fixed missing icons in search fields, including the magnifying glass in <input type="search"> and the filter field icons in Web Inspector. (322219@main) (188297101)

HTML

Resolved Issues

  • Fixed an iframe whose src changed before its first load committed being missing from the back/forward list, which could show the wrong frame content after navigating back. (321768@main) (149977268)
  • Fixed navigating a subframe that still shows its initial about:blank document to replace the history entry instead of adding a new one, so the back button works as expected. (321285@main) (187298226)
  • Fixed changes to the scrolling attribute on an <iframe> not taking effect immediately when overlay scrollbars are in use. (321994@main) (188326221)

Images

Resolved Issues

  • Fixed top-level image viewer to support dark-mode. (322117@main) (97965029)
  • Fixed a broken <img> with an empty alt attribute to have no natural size or aspect ratio, so block-level images no longer show a broken-image icon. (321251@main) (155149701)
  • Fixed the bitmap image naturalWidth and naturalHeight to return the image metadata size rather than the decoded size; this fixed layout of the image for different sizes. (322220@main) (172477431)
  • Fixed image-rendering not applying to generated images, such as cross-fade(), used in the CSS content property. (322533@main) (189024049)
  • Fixed the bitmap image naturalWidth and naturalHeight to return the density-corrected natural rather than the layout size which can be affected the style such as image-orientation. (322534@main) (189024238)

JavaScript

Resolved Issues

  • Fixed BigInt("-") and BigInt("0x ") to throw a SyntaxError. (321129@main) (187423917)
  • Fixed an issue where optimized code could read a stale value of a module variable after an import cycle. (321135@main) (187432418)
  • Fixed spreading or applying a strict mode arguments object truncating its length to 32 bits. (321213@main) (187527803)
  • Fixed stack trace capture to read Error.stackTraceLimit from the Error constructor. (321296@main) (187633441)
  • Fixed named backreferences (\k<name>) inside regular expression lookbehinds incorrectly matching the empty string in some cases. (321355@main) (187741552)
  • Fixed RegExp.prototype.test() skipping the observable read of lastIndex for some anchored regular expressions. (321363@main) (187754780)
  • Fixed ArrayBuffer.prototype.slice() to correctly look up Symbol.species on the constructor, including for ArrayBuffer objects from other realms. (321451@main) (187853060)
  • Fixed Intl.DateTimeFormat.prototype.resolvedOptions() reporting a dayPeriod value for 12-hour formats that was never requested, which caused formatters recreated from those options to output text like “in the morning” instead of “AM”. (321443@main) (187860472)
  • Fixed Intl.Segmenter containing() returning a segment that also included the preceding segment when the index was at a boundary before a surrogate pair, such as an emoji. (322218@main) (187863002)
  • Fixed Intl.DurationFormat dropping zero minutes and seconds that follow a numeric unit. (321755@main) (188252937)
  • Fixed Promise.prototype.then(), catch(), finally(), Promise.resolve(), and Promise.reject() creating promises and errors in the wrong realm when used with a Promise from another realm, such as an iframe. (321864@main) (188345846)
  • Fixed (a?.b)(x) skipping the call and returning undefined when a is null or undefined, instead of evaluating the arguments and throwing a TypeError. (321867@main) (188347937)
  • Fixed Intl.DurationFormat dropping a fractional value folded into a zero unit, so formatting { milliseconds: 500 } with milliseconds: "numeric" returns 0.5 sec instead of an empty string. (321878@main) (188362824)
  • Fixed Intl.DurationFormat dropping the minus sign for fractional values between -1 and 0. (321988@main) (188478180)
  • Fixed String.prototype and RegExp.prototype methods such as match(), replace(), split(), and test() incorrectly running with the caller’s realm when used with a RegExp or function from another realm. (321990@main) (188478512)
  • Fixed tagged templates with a parenthesized optional chain as the tag to keep the correct this value. (322141@main) (188677696)
  • Fixed tagged templates that use a super property as the tag so the tag function is called with the current this value. (322253@main) (188790013)
  • Fixed TypedArray constructors creating an empty array when passed a Proxy that wraps an array-like object. (322260@main) (188794129)
  • Fixed the ArrayBuffer constructor to throw a RangeError for an invalid length before reading the prototype of newTarget. (322281@main) (188822488)
  • Fixed the Array constructor to read the prototype of newTarget before throwing a RangeError for an invalid length. (322282@main) (188822747)

Media

New Features

  • Added support for the transfer option in WebCodecs EncodedAudioChunk, EncodedVideoChunk, AudioData, VideoFrame, and ImageDecoder constructors, allowing ownership of ArrayBuffer data to be transferred instead of copied. (321548@main) (187834056)
  • Added support for the hevc option in VideoEncoder configurations so that WebCodecs can produce HEVC video in Annex B format. (322385@main) (188387670)

Resolved Issues

  • Fixed paused media with audio in a hidden page being able to start playing again without a user gesture. (322266@main) (169468617)
  • Fixed video disappearing while audio continued after returning from Picture-in-Picture to element fullscreen and then exiting to inline on iOS. (321751@main) (181656299)
  • Fixed an issue where text tracks whose mode was set with JavaScript could be overridden by automatic caption selection, preventing multiple text tracks from showing at the same time. (321838@main) (183472153)
  • Fixed an issue where media range responses could drop the last byte of a requested range. (321101@main) (187225949)
  • Fixed a mirrored VideoFrame being flipped vertically instead of horizontally when its rotation is applied. (321697@main) (187300986)
  • Fixed an issue where the play() promise never resolved and the playing event never fired for Media Source Extensions content mixing clear and encrypted media when the decryption key was already available. (321495@main) (187850389)
  • Fixed WebCodecs VideoEncoder and VideoDecoder hanging on flush() when the underlying codec buffers more than four frames. (321682@main) (188165426)
  • Fixed incorrect colors in 8-bit 4:2:2 VP9 video decoded in software, where the chroma channels were swapped and caused a visible color cast. (322477@main) (188857520)
  • Fixed the software VP9 decoder used by WebCodecs and media playback to report a decode error instead of silently dropping 4:4:4 frames or producing corrupted 12-bit frames. (322513@main) (188868857)

Networking

Resolved Issues

  • Fixed cookies with non-ASCII characters set using document.cookie or cookieStore.set() getting truncated, and changed reading them to decode as UTF-8 to match other browsers. (322447@main) (124266311)
  • Fixed resources from app-registered custom URL schemes failing to load because of stricter cross-origin checks. (321782@main) (182251092)
  • Fixed a regression on iOS where Apple Wallet passes delivered through a data: URL navigation did not open in Wallet. (322096@main) (184582548)
  • Fixed an issue where a speculation rules prefetch that hit a cross-origin redirect also cancelled a real navigation to the same URL. (321796@main) (187701208)
  • Fixed an issue where a failed speculative prefetch could break the actual navigation, sometimes prompting users to download the page instead of displaying it. (322091@main) (188119964)
  • Fixed structured header field parsing rejecting the entire header when a dictionary or parameter key begins with *. (321835@main) (188260300)
  • Fixed cross-origin stylesheets loaded without CORS never being reused from the parsed stylesheet cache, improving page load performance. (322418@main) (188872969)

PDF

Resolved Issues

  • Fixed the context menu for selected text in PDFs on macOS not offering the Translate option. (322064@main) (186553741)

Performance

Resolved Issues

  • Fixed unnecessary style resolution for ::before and ::after rules that cannot generate content, such as box-sizing resets, improving performance. (322517@main) (186807298)
  • Fixed synthetically bolded text always using the slower text measurement path. (321394@main) (187588326)
  • Fixed a regression where items with a percentage flex-basis inside deeply nested column flex containers could make layout time grow exponentially and freeze the page. (321866@main) (187727635)

Rendering

New Features

  • Added live pinch-to-zoom on macOS, so content stays sharp during the gesture and fixed and sticky elements follow the zoom. (322535@main) (106728296)

Resolved Issues

  • Fixed synthetic bold text being measured wider than the font’s real metrics, which caused letter-spaced text to overflow and wrap and skewed ch units and tab-size. (321206@main) (55307109)
  • Fixed a retriggered overlay scrollbar expansion animation jumping instead of animating smoothly from its current progress. (321389@main) (183216907)
  • Fixed the intrinsic width of replaced elements with a percentage width and a resolvable percentage height. (321748@main) (187522214)
  • Fixed floats and inline-blocks being sized incorrectly when a child sets min-width: min-content or min-width: max-content. (321332@main) (187621931)
  • Fixed hit testing on 3D-transformed elements rotated so far that part of them is behind the viewer, where the visible portion did not receive pointer events. (321633@main) (188120032)
  • Fixed an issue where white-space: nowrap content containing spaces overflowed the line instead of wrapping to the next line. (322209@main) (188406135)
  • Fixed an issue where inline box borders were drawn closed on a fragment where the inline box continues past a line-clamp clamp point. (322345@main) (188471367)
  • Fixed unnecessary rendering and compositing work when a frame moves without resizing, improving scrolling performance on pages with fixed-position iframes. (322194@main) (188665149)

SVG

Resolved Issues

  • Fixed calling clear() on a pattern’s patternTransform.baseVal so the patternTransform attribute is ignored. (321124@main) (170518953)
  • Fixed getComputedTextLength() and the character position methods to account for lengthAdjust="spacingAndGlyphs". (321445@main) (187473448)
  • Fixed SVG angle values to serialize numbers using CSS number formatting, consistent with SVG lengths and CSS values. (321460@main) (187757273)
  • Fixed a regression where an SVG without a viewBox had zero width as a flex item. (321446@main) (187794543)
  • Fixed the transform attribute on the outermost <svg> element being ignored or applied in the wrong position. (321441@main) (187854814)
  • Fixed pointer-events values stroke, visibleStroke, visible, and all missing the stroke area of SVG shapes when stroke is none. (322143@main) (188285211)
  • Fixed paused SVG image animations restarting from the beginning instead of resuming, matching the behavior of animated bitmap images. (322303@main) (188834229)

Security

Resolved Issues

  • Fixed navigator.hardwareConcurrency returning unusual core counts to known fingerprinting scripts, which could trigger bot detection challenges, by alternating between 4 and 8 instead. (321431@main) (177211562)
  • Fixed Content Security Policy violations in workers not firing securitypolicyviolation events or notifying a ReportingObserver. (322099@main) (188368580)

Web API

Resolved Issues

  • Fixed Navigation API dispose events for entries removed by another frame’s navigation seeing stale navigation.entries(), and fixed a crash when a frame is detached during a dispose event. (321726@main) (184961466)
  • Fixed an issue where ending a worker with both a readwrite and a readonly IndexedDB transaction in flight could leave that database permanently unable to run any transaction. (321564@main) (187349025)
  • Fixed window.outerWidth and window.outerHeight returning 0 in some apps that embed WebKit. (321417@main) (187603389)
  • Fixed pagehide, visibilitychange, and unload event handlers having no effect, such as beacons and messages never arriving, when navigating to a cross-site page that loads in a different process. (321626@main) (188113419)
  • Fixed memory growth when reading Blob data, such as with Blob.arrayBuffer() or FileReader, by no longer storing completed reads in the memory cache. (321948@main) (188322145)
  • Fixed an issue where calling preventDefault() in a navigate event handler did not cancel a cross-site navigation.navigate(). (322591@main) (188636761)
  • Fixed Largest Contentful Paint to use an image’s natural dimensions when computing its effective visual area, so SVG images with only one natural dimension are no longer reported as candidates. (322624@main) (189112402)

Web Extensions

Resolved Issues

  • Fixed webRequest events not firing for subframes loaded inside an extension’s action popup. (321290@main) (167403221)
  • Fixed an issue where a web extension’s background service worker could handle messages before it finished registering after being reloaded, causing checks for existing offscreen documents to fail. (321556@main) (186304610)
  • Fixed webRequest event listeners filtered by types so that main_frame and sub_frame match top-level and iframe loads respectively instead of being swapped. (321074@main) (187214709)
  • Fixed webRequest events reporting the wrong parentFrameId for subframes whose parent is the main frame, which now correctly report 0. (321079@main) (187227773)
  • Fixed an issue where HTTP and HTTPS iframes loaded inside a web extension page incorrectly received the extension’s Content Security Policy. (322449@main) (188407446)
  • Fixed an issue where an extension offscreen document calling window.close() did not close the document. (321974@main) (188408957)

Web Inspector

Resolved Issues

  • Fixed very large values, such as border-radius: calc(infinity * 1px), overflowing the box model diagram. (322462@main) (174246956)
  • Fixed a regression where the Sources tab showed no scripts when inspecting an older remote target. (322286@main) (187134944)
  • Fixed JavaScript breakpoints showing an incorrect resolved state on pages with multiple frames when one frame navigates or reloads. (322410@main) (187726262)
  • Fixed exported Timeline recordings losing the DOM node associated with layout and rendering events when imported again. (321679@main) (188164899)

WebAssembly

New Features

  • Added support for WebAssembly wide arithmetic, which adds instructions for 128-bit integer addition and subtraction and for multiplying two 64-bit integers into a 128-bit result. (321926@main) (188342106)

Resolved Issues

  • Fixed an inefficiency in the WebAssembly in-place interpreter by making its instruction handlers smaller, improving performance by about 5% on Apple silicon. (321217@main) (187291986)
  • Fixed redundant work when compiling WebAssembly code that uses garbage-collected structs and arrays, improving compile times. (322019@main) (188134232)

WebDriver

New Features

  • Added initial support for the WebDriver BiDi browsingContext.captureScreenshot command, which captures the page viewport as a PNG. (321353@main) (145987196)

Resolved Issues

  • Fixed WebDriver key actions so that the Insert, Pause, and Cancel keys report the correct DOM key and code values, and so that keypad keys on iOS report Numpad* codes and the numpad key location. (322405@main) (187036197)
  • Fixed WebDriver to queue classic commands for a session so they run one at a time, preventing hangs when commands overlap. (322485@main) (187215746)
  • Fixed an issue where a WebDriver command waiting for navigation to complete could hang indefinitely when another command on the same page also started waiting, by returning an error for the replaced command. (321165@main) (187217320)
  • Fixed returning a stale element from a script to report a “stale element reference” error instead of “no such element”. (322221@main) (187510557)

WebGL

Resolved Issues

  • Fixed a regression where WebGL texImage2D() and texSubImage2D() uploads from image sources handled premultiplied alpha incorrectly. (321143@main) (187093618)

WebGPU

Resolved Issues

  • Fixed copyExternalImageToTexture() to preserve alpha values instead of passing lossy premultiplied data, which caused some textures with transparency to render incorrectly. (321601@main) (142087137)
  • Fixed createComputePipelineAsync() and createRenderPipelineAsync() never settling when the device was lost or destroyed. (321466@main) (187846864)
  • Fixed fragment shaders that discard before loops writing NaN values to the render target on some Apple silicon GPUs. (321870@main) (187997601)
  • Fixed WebGPU content permanently dropping to a fraction of the display refresh rate after a few slow frames. (322439@main) (188257377)
  • Fixed an issue where render pipelines created from an already-used GPUShaderModule always missed the shader cache, making repeated pipeline creation as slow as a cold compile. (321998@main) (188501171)
  • Fixed an issue where reusing a render bundle with executeBundles() lost the bind group dynamic offsets after the first execution. (322002@main) (188524116)
  • Fixed setBindGroup() on render pass encoders doing redundant sampler registration work. (322005@main) (188530801)

WebRTC

Resolved Issues

  • Fixed H.265 video encoded or decoded through WebRTC and WebCodecs always being tagged as BT.709 limited range instead of preserving its actual color space. (321788@main) (187787801)
  • Fixed full-range color handling for H.264 video in WebRTC and WebCodecs by decoding it with WebKit’s own VideoToolbox decoder. (322034@main) (188171149)
  • Fixed H.264 video encoding and decoding to preserve the video’s color space, and fixed frames from a canvas to use the sRGB color space when encoded. (322056@main) (188373969)

October 08, 2026 11:29 PM

October 06, 2026

WPE WebKit Blog: WPEPlatform: the new WPE API

Igalia WebKit

With the 2.54 release, WPEPlatform became the default way of integrating WPE WebKit with the underlying platform, and its API is now considered stable. At the same time, the libwpe-based API that WPE has relied on since its beginnings is now deprecated. In this article we explain why this change was made, how applications use the new API, and what is involved in migrating an existing application from libwpe.

Back in 2022 we published an overview of the WPE WebKit project, describing the different components of WPE and how they fit together. Most of what that article explains about WebKit and the WPE API is still valid, but the way WPE connects to the platform has changed considerably since then, so this is a good moment to revisit that part of the picture.

How WPE used to integrate with the platform

Since its beginnings, WPE has delegated graphics and input to code that lives outside of WebKit. libwpe defined a generic interface for these, and a backend, loaded at runtime, implemented that interface for a given platform. WPEBackend-fdo, based on Wayland and other freedesktop.org technologies, was the reference backend, while Cog provided a convenience layer on top for those writing browsers.

This design achieved its main goal: hardware vendors and integrators could support WPE on their platforms without having to modify WebKit itself. However, it also meant that a good part of the platform work was left to the application. To show a web page on screen using WPEBackend-fdo, an application had to:

  • Load a libwpe backend with wpe_loader_init() and initialize it, usually by passing it an EGL display.
  • Create a view backend through the “exportable” API of WPEBackend-fdo, wrap it in a WebKitWebViewBackend, and pass that to the web view.
  • Receive, in a set of callbacks, the buffers exported by WebKit for each frame, present them on screen, release them, and notify WebKit when the frame had been displayed.
  • Collect input events from the windowing system and forward them to WebKit with the wpe_view_backend_dispatch_*_event() family of functions.

As a result, even a simple browser required a fair amount of knowledge of the graphics pipeline, and the code involved was split across three repositories (WebKit, libwpe, and WPEBackend-fdo) that had to be kept in sync. Much of this code was the same from one application to the next, and Cog provided a ready-made implementation of it for applications that did not want to write their own.

What WPEPlatform changes

WPEPlatform is a GObject-based library that lives in the WebKit repository and is built as part of WPE WebKit. It replaces both libwpe and the backends written against it, and it moves the responsibility of rendering and input handling away from the application and into WebKit and the platform implementation.

The library is only used by WebKit’s UI process. As before, the web process renders the page into buffers that are shared with the UI process (DMA-BUF buffers on Linux, AHardwareBuffer on Android, or shared memory when rendering in software). The difference is that WebKit now hands these buffers directly to the platform implementation, which takes care of presenting them, without the application having to take part in the process.

The API is built around four classes:

  • WPEDisplay: the connection to the platform, and the object that creates all the others.
  • WPEToplevel: a top-level surface, which is normally a window, or the whole output on platforms that have no windows.
  • WPEView: the surface where a single web view is rendered, hosted in a toplevel. It also receives the input events for that web view.
  • WPEBuffer: the pixel data produced by WebKit and handed to the view, with a subclass for each type of buffer.

These are complemented by smaller classes for screens, keymaps, settings, input methods, gestures, gamepads, the clipboard, and accessibility. Several of them, such as screens, settings, input methods, and gestures, had no equivalent in libwpe.

WPE WebKit ships with three built-in platform implementations: Wayland, DRM/KMS (for devices that drive the display directly, without a compositor), and headless (for testing and offscreen rendering). Other implementations can be developed out of tree and installed as modules, which WPE WebKit discovers at runtime. wpe-platform-gtk, which embeds WPE web views in GTK 4 applications, is an example of this.

A diagram of the WPE architecture: the application uses WPE WebKit, which includes the WebKit engine and API and the WPEPlatform API. WPEPlatform has built-in Wayland, DRM/KMS, and headless implementations, and custom implementations can be provided for other platforms.

Using WPEPlatform in an application

For most applications, WPEPlatform is not visible at all. A WebKitWebView created without a backend selects a platform automatically: WPE WebKit tries the available platform implementations in order of priority and uses the first one that connects successfully. The following program is a complete, if minimal, browser:

#include <wpe/webkit.h>
#include <wpe/wpe-platform.h>

static void
on_view_closed (WPEView *view, gpointer user_data)
{
    g_main_loop_quit (user_data);
}

int
main (int argc, char *argv[])
{
    g_autoptr(GMainLoop) loop = g_main_loop_new (NULL, FALSE);
    g_autoptr(WebKitWebView) web_view = g_object_new (WEBKIT_TYPE_WEB_VIEW, NULL);

    WPEView *view = webkit_web_view_get_wpe_view (web_view);
    g_signal_connect (view, "closed", G_CALLBACK (on_view_closed), loop);

    webkit_web_view_load_uri (web_view, argc > 1 ? argv[1] : "https://wpewebkit.org");
    g_main_loop_run (loop);

    return 0;
}

It only needs the wpe-webkit-2.0 pkg-config module to build, which pulls in wpe-platform-2.0 when WPE WebKit is built with WPEPlatform support (the default since 2.54). Note that the closed signal is emitted when the user closes the window, and of the built-in platforms, only Wayland emits it. On DRM and headless the application decides by itself when to exit.

The platform API comes into play when an application needs more than these defaults. The most common cases are the following.

Choosing a platform. The WPE_PLATFORM environment variable (wayland, drm, or headless) forces a given platform without any changes to the code. An application that is meant to run on a single platform can also create the display itself and pass it to the web view:

#include <wpe/wayland/wpe-wayland.h>

g_autoptr(GError) error = NULL;
g_autoptr(WPEDisplay) display = wpe_display_wayland_new ();
if (!wpe_display_connect (display, &error))
    g_error ("Could not connect to Wayland: %s", error->message);

WebKitWebView *web_view = g_object_new (WEBKIT_TYPE_WEB_VIEW,
                                        "display", display,
                                        NULL);

The Wayland-specific API is available with the same wpe-webkit-2.0 module, but an application that depends on it can check for the wpe-platform-wayland-2.0 pkg-config module, which is only installed when WPE WebKit is built with the Wayland platform enabled. The DRM and headless implementations work the same way, with wpe_display_drm_new() and wpe_display_headless_new().

Handling input. Input events are delivered to the WPEView through the event signal before they reach the web page. An application can connect to it to implement keyboard shortcuts, returning TRUE to stop the event from reaching the page:

static gboolean
on_view_event (WPEView *view, WPEEvent *event, gpointer user_data)
{
    if (wpe_event_get_event_type (event) == WPE_EVENT_KEYBOARD_KEY_DOWN
        && (wpe_event_get_modifiers (event) & WPE_MODIFIER_KEYBOARD_CONTROL)
        && wpe_event_keyboard_get_keyval (event) == WPE_KEY_q) {
        g_main_loop_quit (user_data);
        return TRUE;
    }
    return FALSE;
}

g_signal_connect (view, "event", G_CALLBACK (on_view_event), loop);

Controlling the window. The toplevel that hosts the view is available through wpe_view_get_toplevel(), and provides functions to set the title (wpe_toplevel_set_title()), resize, maximize, or switch to fullscreen (wpe_toplevel_fullscreen()). Changes in its state are reported by the toplevel-state-changed signal of the view. How much of this is supported depends on the platform: Wayland implements all of it, the headless implementation only keeps track of size and fullscreen state, and DRM does not support them, since its toplevel always covers the whole screen.

Finally, wpe_display_get_settings() gives access to WPESettings, where applications can override platform settings such as the dark mode preference, reduced motion, or contrast. The prefers-color-scheme, prefers-reduced-motion, and prefers-contrast CSS media queries follow these settings.

Migrating from libwpe

For an application, most of the migration consists of removing code. With WPEBackend-fdo, creating a web view looked roughly like this, with most of the actual work happening in the callbacks of exportable_client:

struct wpe_view_backend_exportable_fdo *exportable =
    wpe_view_backend_exportable_fdo_egl_create (&exportable_client, app, width, height);
struct wpe_view_backend *wpe_backend =
    wpe_view_backend_exportable_fdo_get_view_backend (exportable);
WebKitWebViewBackend *backend =
    webkit_web_view_backend_new (wpe_backend, destroy_backend, exportable);
WebKitWebView *web_view = webkit_web_view_new (backend);

With WPEPlatform, the same is achieved with a single line:

WebKitWebView *web_view = g_object_new (WEBKIT_TYPE_WEB_VIEW, NULL);

webkit_web_view_new() is only available when WPE WebKit is built with the legacy API, so applications need to switch to g_object_new(). The rest of the application code that uses the WebKit API (settings, network sessions, navigation, and so on) does not need any changes. The diagram below summarizes what moves out of the application:

A diagram comparing the two architectures. Before: the application, often through Cog, uses WPE WebKit, libwpe, and WPEBackend-fdo, and takes care of loading the backend, handling exported buffers, and dispatching input. After: the application only uses the WebKit API, and WPEPlatform, inside WPE WebKit, handles rendering and input.

The rest of the libwpe code in an application maps to WPEPlatform as follows:

libwpe / WPEBackend-fdo WPEPlatform
wpe_loader_init() and backend initialization Automatic, or WPE_PLATFORM, or creating a WPEDisplay
Exportable client callbacks, buffer release, frame completion Handled by WebKit and the platform implementation
wpe_view_backend_dispatch_*_event() Handled by the platform; intercept with the WPEView::event signal
Fullscreen handler wpe_toplevel_fullscreen() and the toplevel state

There are a few things in libwpe and WPEBackend-fdo that have no equivalent in WPEPlatform. The process provider API of libwpe is gone, since WebKit is again in charge of launching its auxiliary processes. The only exception is Android, which has its own WPEProcessManager API. The audio extension of WPEBackend-fdo has not been supported since 2.46, and the video plane extension, used to display video frames on hardware overlay planes, has no equivalent yet.

Those who maintain a custom backend for WPEBackend-fdo, rather than an application, will need to port it to a WPEPlatform implementation. This means subclassing WPEDisplay, WPEToplevel, and WPEView, and optionally the input method, keymap, and screen classes. The tutorial on writing a platform implementation covers this in detail. In many cases, extending one of the built-in implementations might be enough: the Wayland one, for example, exposes its wl_display, wl_compositor, and wl_surface objects, which makes it possible to add support for additional Wayland protocols.

On the build side, the legacy API is still built by default in 2.54 and will continue to be maintained while applications migrate. Once an application no longer needs it, it can be disabled at build time with the ENABLE_WPE_LEGACY_API=OFF CMake option. As for Cog, it will not have further stable releases beyond the 0.18.x series, so we recommend that new applications use the WebKit and WPEPlatform APIs directly.

Further reading

The reference documentation for WPEPlatform has been considerably extended for this release, and is the best place to continue from here:

For an example of a platform implementation that lives entirely outside of WebKit, Alex has written about WPE WebKit on Android with WPEPlatform, which shows how far the API can be taken.

If you run into problems while migrating, or find that something you relied on in libwpe is missing, please let us know, either by filing a bug in WebKit’s Bugzilla, on the webkit-wpe mailing list, or in the #wpe Matrix channel.

October 06, 2026 12:00 AM

October 05, 2026

Igalia WebKit Team: WebKit Igalia Periodical #80

Igalia WebKit

Update on what happened in WebKit in the week from September 28 to October 5.

Another week populated with updates! This time we sweep through a new crypto backend, more layer-based SVG engine updates, better filters, and some miscellaneous updates.

Cross-Port 🐱

Add moving steps handling to map, img, label and radio input elements to handle their internal book keeping. This also handles custom element registries.

Initial support to build WPE and WebKitGTK with the OpenSSL crypto backend has landed. This is still in development and not yet intended for use, but progress on bringing it to parity with the libgcrypt backend is ongoing.

Graphics 🖼️

Fixed stacked backdrop filters in the Skia compositor, so an element with a backdrop-filter placed over an earlier element with its own backdrop-filter now shows that earlier element with its filter applied, instead of without it.

Removed redundant save/restore pairs when painting transformed SVG shapes in the Layer-Based SVG Engine (LBSE), so each shape now uses a single pair, just like the legacy SVG engine. When DOM rendering happens in the GPU process, every state change is an IPC message, so this cuts LBSE from about 8.5 to about 6.5 messages per shape and closes the gap to the legacy engine.

Fixed transformed <clipPath> elements that use clipPathUnits="objectBoundingBox" in the Layer-Based SVG Engine (LBSE), whose transform was wrongly applied in bounding box units instead of user space. Rendering and hit testing now apply it in user space, matching the legacy SVG engine.

Fixed hit testing and repainting of non-scaling strokes in the Layer-Based SVG Engine (LBSE), which kept using the old stroke width after the transform of the shape or one of its ancestors changed. A vector-effect: non-scaling-stroke stroke depends on every transform up to the outermost <svg> element, so its cached bounds are now cleared whenever one of those transforms changes.

Fixed the source area of filters and backdrop filters: a filter's offscreen surface only covered the visible part of its layer, so when a layer was partly scrolled out of the viewport, its drop shadow was cut off even where the shadow itself fell inside the viewport. The surface now also covers the area the filter reads from outside the visible part, as in Chromium, and a backdrop-filter blur is mirrored at the border box as the spec requires, which fixes several web-platform-tests.

Fixed viewport clipping of nested <svg> elements with overflow: hidden in the Layer-Based SVG Engine (LBSE), which did not clip descendants that have their own layer, such as a transformed <g>. Containers that clip their overflow now get their own layer, like containers using clip-path or mask already do, so layered descendants are clipped correctly.

Fixed misplaced child layers of SVG elements with a filter in the Layer-Based SVG Engine (LBSE), which were drawn in the wrong position whenever an ancestor layer had a transform applied.

WPE WebKit 📟

Restored the Direct Rendering Manager (DRM) vertical blank monitor for the legacy WPE API, so rendering is once again paced by the display's actual refresh cycle instead of a fixed 60 Hz timer. Two regressions had silently disabled it: a missing display ID forced the timer fallback, and a shadowed variable broke the display controller lookup.

That’s all for this week!

By Igalia WebKit Team at October 05, 2026 07:21 PM

September 29, 2026

Igalia WebKit Team: WebKit Igalia Periodical #79

Igalia WebKit

Update on what happened in WebKit in the week from September 21 to September 28.

Following the stable releases last week, we have the first security advisory covering fixes included in them; the flurry of fixes and polish for the Skia compositor continues; SharedArrayBuffer gets re-enabled, and more.

Cross-Port 🐱

Fixed a crash on exit in WebKitGTK and WPE WebKit, caused by the EGL display being torn down in an atexit() handler while other threads were still using it, by now destroying it on the main thread once the GPU process has stopped its WebGL thread and the web process has closed its pages.

Fixed moving steps for source element to account for the <media> and <model> elements.

JavaScriptCore 🐟

The built-in JavaScript/ECMAScript engine for WebKit, also known as JSC or SquirrelFish.

SharedArrayBuffer is now enabled for the WPE and GTK ports. SharedArrayBuffer is exposed when the origin is cross-origin isolated. We have enabled 35 associated cross-origin isolation tests. Initialization was designed to work with Android service-launched processes, too.

Graphics 🖼️

Fixed a bug with 3D-transformed elements that are tilted so far away from the viewer that part of them ends up behind the camera, where WebKit got the visible area of the element wrong and computed the hidden part instead. As a result, the visible part of such elements is no longer left blank in WebKitGTK and WPE—painting and hit-testing works properly now.

Fixed oversized damage regions for 3D-transformed layers that cross the camera plane, by clipping their projected polygon against the clip bounds so the damage now matches what the layer actually paints.

GPU renderer small paths now use Skia distance fields. Bitmap atlas entries depended on the transform and subpixel position, causing animated paths to be re-rasterized and uploaded every frame. With distance field indexed entries we can reuse them across transforms. We are improving MotionMark Suits results on Raspberry Pi 4 and desktop.

Fixed the paint order of preserve-3d layers that cross the camera plane in the Skia-based compositor. Layer corners behind the camera used to be flipped to the far side of the screen when projected—layers were sorted against the wrong geometry and far layers painted over near ones. Now the part of a layer that lies behind the camera is clipped away before sorting, which makes complex 3D scenes like the CSS FPS demo render correctly.

Aligned the filter surface with the device pixel grid in the Skia-based compositor, so layers with CSS filters that end up at fractional device positions are no longer resampled when the filtered result is drawn. This fixes wrong colors in the outermost rows and columns of pixels of such layers, which now render the same as they would without a filter.

Fixed SMIL-animated viewBox changes being ignored on the outermost <svg> element in the Layer-Based SVG Engine (LBSE), bringing it in line with the legacy SVG engine.

Fixed edge artifacts on scaled or transformed non-repeating background images in the Skia-based compositor, which painted all composited background images as if they repeat, so a thin line of the opposite edge showed up along the border. Only tiles that actually repeat are painted that way now, which removes the yellow line that showed up above the gun in the CSS FPS demo. This fixes the last known artifact in that demo.

Fixed rendering artifacts in WebKitGTK and WPE WebKit for 3D-transformed layers that extend behind the camera under a perspective (for example a large element under rotateX(90deg)), whose re-composited area came out too small because corners behind the camera were projected onto the wrong side of the frame, and are now clipped away before mapping.

WebKitGTK 🖥️

Fixed touch positions in the GTK port, which were offset by the surface transform, the room left for the client-side decoration shadows, because it was not subtracted from the raw GdkEvent position, so every touch landed away from the finger unless the window was maximized. The same change cancels the touch sequences that were stranded when the web view is unmapped or a dialog was shown, which left the page with a finger that was never released.

WPE WebKit 📟

WPE Platform API 🧩

New, modern platform API that supersedes usage of libwpe and WPE backends.

The ENABLE_WPE_PLATFORM build option has been removed. The new WPE Platform API, which was enabled by default since 2.54, is now always enabled.

Releases 📦️

A new security advisory, WSA-2026-0006, has been published for WebKitGTK and for WPE. This covers issues fixed in the recent 2.54.0 releases, and for the first time includes security issues fixed in the bundled ANGLE and Skia libraries. Updating to the latest stable release is highly recommended.

That’s all for this week!

By Igalia WebKit Team at September 29, 2026 12:40 AM

September 23, 2026

Release Notes for Safari Technology Preview 253

Surfin’ Safari

Safari Technology Preview Release 253 is now available for download for macOS Golden Gate and macOS Tahoe. If you already have Safari Technology Preview installed, you can update it in System Settings under General → Software Update.

This release includes WebKit changes between: 320113@main…321067@main.

Accessibility

Resolved Issues

  • Fixed an issue where VoiceOver announced aria-keyshortcuts values containing Meta or Alt literally instead of using the macOS terminology Command and Option. (320424@main) (186342070)
  • Fixed an issue where VoiceOver could repeatedly announce the same live region content as it streamed into a page. (320576@main) (186694747)

Animations

New Features

  • Added support for style-originated scroll timelines to match globally, allowing them to be defined outside of the target’s hierarchy or that of an element with a timeline-scope property. (320183@main) (186261285)

Resolved Issues

  • Fixed an issue where an animation attached to a view timeline’s scroll range was not updated when the scroll container’s scrollable overflow changed. (320244@main) (185328465)
  • Fixed an issue where an animation could attach to a style-originated timeline made visible by timeline-scope instead of one established by an ancestor, which now takes priority. (320223@main) (186265325)
  • Fixed an issue where a style-originated timeline defined outside of a timeline-scope hierarchy could remain active instead of yielding an inactive timeline. (320227@main) (186330965)
  • Fixed an issue where changing a timeline-scope value did not update timelines for animations outside of its hierarchy. (320231@main) (186331685)
  • Fixed a regression where a paused and seeked animation incorrectly finished when resumed after a separate animation running at a non-default playback rate. (320238@main) (186339093)
  • Fixed an issue where the ViewTimeline constructor did not require a subject parameter. (320969@main) (187200265)

CSS

New Features

  • Added support for the extended numeric range in longhand East Asian counter styles. (320887@main) (109875198)
  • Added support for CSSContainerRule.conditions. (320762@main) (182257864)
  • Allow combining safe and unsafe keywords with normal alignment, and change safe behavior for absolutely (and fixed positioned) boxes to keep the box within their original containing block (the viewport). (185952139)
  • Added support for using sibling-index() and sibling-count() within container query conditions. (320493@main) (186564084)

Resolved Issues

  • Fixed an issue where a near-zero fixed background-size value collapsed the image tile to nothing. (320169@main) (140387662)
  • Fixed an issue where the [class~=foo] attribute selector did not perform as well as an equivalent class selector. (320862@main) (164128575)
  • Fixed an issue where highlight colors, such as those used by ::selection, did not inherit as a StyleColor, which could prevent values like color-mix(in oklab, teal 50%, currentcolor) from resolving correctly. (320228@main) (184495338)
  • Fixed an issue where a second CSS custom highlight sharing a Range with an already-registered highlight never painted. (320355@main) (185173794)
  • Fixed an issue where outside list markers were only repositioned after layout when their first formatted line was in a descendant block, instead of for every outside list marker. (320561@main) (185529863)
  • Fixed an issue where an inset box-shadow with a large spread on a wrapped inline element painted outside the element as full-width bands. (320150@main) (185651318)
  • Fixed an issue where CSS.highlights iterated in hash order instead of registration order after its wrapper was garbage collected. (320343@main) (185754041)
  • Fixed an issue where corner-shape rendered incorrectly when inner corners intersected. (320287@main) (185931169)
  • Fixed an issue where a grid item with an aspect-ratio could keep a stale inline size contribution when the grid container shrank. (320117@main) (186101273)
  • Fixed an issue where flexible grid tracks did not respect the grid container’s minimum and maximum size. (320160@main) (186103325)
  • Fixed an issue where fixed grid track sizing functions were overridden to zero while sizing tracks to fit non-spanning items. (320168@main) (186117142)
  • Fixed an issue where explicitly-placed grid items advanced the auto-placement cursor. (320253@main) (186118858)
  • Fixed an issue where the propagated root background was painted in the wrong position in vertical-rl writing mode. (320167@main) (186274088)
  • Fixed an issue where opening a <details> element made its <summary> one pixel shorter. (320461@main) (186413478)
  • Fixed an issue where a typed parameter of a CSS custom function did not keep its type. (320340@main) (186439114)
  • Fixed an issue where attr() stopped invalidating on attribute changes after a view transition. (320420@main) (186467515)
  • Fixed an issue where unicode-bidi and direction had no effect on an inside ::marker. (320504@main) (186468752)
  • Fixed an issue where fit-tolerance did not interpolate between <length-percentage> values. (320383@main) (186509338)
  • Fixed an issue where a grid item’s automatic minimum size was not clamped to a fixed maximum track sizing function. (320750@main) (186606479)
  • Fixed a regression where the line-height quirk in quirks mode was not applied to line boxes inside a nested inline-block element that had no line-height of its own. (320514@main) (186614663)
  • Fixed an issue where interpolating a <length-percentage> from a calc() value to a pure <length> dropped the percentage component at 100% progress. (320585@main) (186628152)
  • Fixed an issue where text-decoration-thickness and text-underline-offset did not preserve percentage values when interpolating. (320584@main) (186628810)
  • Fixed an issue where align-content left the list marker behind. (320532@main) (186676835)
  • Fixed an issue where text-indent moved an outside list marker along with the indented text. (320533@main) (186678552)
  • Fixed an issue where a float in a list item’s content pushed the outside list marker inward with the line. (320549@main) (186680551)
  • Fixed an issue where a tab character rendered too narrow when tab-size was small in a proportional font. (320671@main) (186698040)
  • Fixed an issue where a gradient in the content property painted blank. (320620@main) (186742805)
  • Fixed an issue where some combinations of corner-shape with thick borders rendered incorrectly. (320826@main) (186808871)
  • Fixed an issue where fit-content() grid tracks could be stretched beyond their argument instead of capping the track’s growth limit. (320780@main) (186961437)
  • Fixed an issue where an orthogonal <caption>‘s margins were missing from the table. (321056@main) (187137359)
  • Fixed an issue where an orthogonal <caption> ignored its margin against the table edge. (321060@main) (187139068)
  • Fixed an issue where transforming a table row group could misplace its absolutely positioned children. (321065@main) (187252945)
  • Fixed an issue where the resolved right and bottom values of an out-of-flow positioned element were wrong inside a vertical inline containing block. (321061@main) (187295559)
  • Fixed an issue where the identity and translation fast path of a transformation matrix ignored the w component when mapping a 4-component point. (321030@main) (187297446)

Canvas

Resolved Issues

  • Fixed an issue where a placeholder <canvas> element with no pushed OffscreenCanvas frame would fail instead of returning a transparent black image. (320202@main) (186229685)
  • Fixed an issue where an off-by-one error in the bottom-row check caused canvas noise injection to misclassify the bottom-left pixel. (320510@main) (186640624)
  • Fixed a performance regression where drawing a canvas onto itself with drawImage prevented its backing surface from being recycled. (320735@main) (186731221)
  • Fixed an issue where a canvas 2D context remained unusable after a temporary failure to allocate its backing store. (320605@main) (186799597)
  • Fixed an issue where drawing to <canvas> computed path bounds unnecessarily when the whole backing store was already marked dirty. (320709@main) (186946552)

Editing

Resolved Issues

  • Fixed an issue where the context menu in PDFs with copying disabled was missing text selection options. (320444@main) (186527750)
  • Fixed an issue on macOS where the Copy option was enabled in the edit menu after selecting text in a PDF that disallows copying. (320534@main) (186648332)

Forms

Resolved Issues

  • Fixed an issue where a large picker for a base-appearance <select> could render outside the viewport, by applying safe alignment to keep it within its original containing block. (320564@main) (185952139)

HTML

Resolved Issues

  • Fixed an issue where the window load event could fail to fire if a readystatechange handler triggered a new load during page completion. (320331@main) (186373617)

JavaScript

New Features

  • Added support for BigInt values in Intl.PluralRules.prototype.select and Intl.PluralRules.prototype.selectRange. (320396@main) (186534585)
  • Added support for a faster Toom-3 multiplication algorithm for large BigInt values. (321013@main) (187336927)

Resolved Issues

  • Fixed an issue in JavaScriptCore where a stale inline cache for a custom accessor on a previously flattened dictionary could persist after the accessor was shadowed, which could cause code that replaces built-in properties at runtime (such as a test mocking library overriding XMLHttpRequest) to keep using the original value. (320742@main) (180048596)
  • Fixed an issue where Intl.DurationFormat in digital style included a stray separator when minutesDisplay was set to "auto" and minutes ended up hidden. (320182@main) (180722012)
  • Fixed an issue where Array.prototype.toSpliced threw a TypeError instead of a RangeError when the array’s length was Infinity. (320554@main) (184438837)
  • Fixed an issue where TypedArray setFromTypedArray could not use memmove when the region was intentionally overlapping and the spec algorithm needed to read back the modified result. (320130@main) (186145219)
  • Fixed an issue where the TypedArray constructor could produce incorrect results when copying Array content whose element access has side effects. (320185@main) (186227333)
  • Fixed a performance issue where String.prototype.toLowerCase and String.prototype.toUpperCase did not scan strings inline in the DFG and FTL JIT tiers. (320293@main) (186301688)
  • Fixed an issue where a character following a class set operand in a /v mode regular expression class incorrectly added U+0000 to the class. (320216@main) (186319779)
  • Fixed an issue where Temporal.ZonedDateTime.prototype.round resolved the rounded wall-clock time using a minute-rounded offset instead of the correct offset. (320217@main) (186319834)
  • Fixed an issue where Object.freeze() did not invalidate the megamorphic inline cache epoch, which could cause stale property accesses on a frozen object. (320276@main) (186393610)
  • Fixed an issue where deleting a property of a dictionary prototype in place did not invalidate the megamorphic store cache. (320375@main) (186513123)
  • Fixed a performance issue where RegExp.prototype.test did not fast-fail when the input string was shorter than the pattern’s minimum possible match length. (320487@main) (186534079)
  • Fixed an issue where regular expression lookbehind assertions could fail to match, or match at an invalid position, due to incorrect handling of surrogate pairs in the regular expression interpreter. (320492@main) (186536137)
  • Fixed an issue where Atomics.isLockFree() converted its argument with a 32-bit integer conversion instead of ToIntegerOrInfinity, which could cause it to incorrectly return true for large size values. (320715@main) (186610789)
  • Fixed an issue where a stack frame for a script loaded from a data: URL included the entire script in Error.stack instead of a truncated URL. (320770@main) (186614179)
  • Fixed an issue where a JIT fast path for RegExp.prototype.test could skip the required read of lastIndex, silently dropping observable side effects. (320582@main) (186792229)
  • Fixed an issue where tail-call optimization was skipped for eval calls that didn’t resolve to the real eval function, and incorrectly applied inside generator and async function bodies, which could produce incorrect results. (320687@main) (186840477)
  • Fixed an issue where RegExp.escape incorrectly narrowed supplementary code points to 16 bits. (320682@main) (186962453)
  • Fixed an issue where a \- following a class set operand in a /v mode regular expression class incorrectly threw a SyntaxError. (320683@main) (186962513)
  • Fixed an issue where RegExp::deleteCode() cleared a regular expression’s cached pattern atom, which could cause incorrect values from static RegExp properties such as leftContext and rightContext after the compiled code was reclaimed while idle. (320817@main) (187104089)
  • Fixed an issue where String.prototype.at and String.prototype.codePointAt could return an incorrect value for an out-of-bounds or negative index because the JIT compiler could eliminate their bounds check. (321014@main) (187336661)

Media

Resolved Issues

  • Fixed an issue where wireless playback could create a remote media session helper too eagerly, which could delay switching an active playback route to a wireless device. (320748@main) (184536781)
  • Fixed an issue where MediaRecorder could hold back a lone video keyframe indefinitely instead of using it to start a new recording segment. (320219@main) (186229843)
  • Fixed an issue where an AV1 sequence header that exactly filled the buffer was incorrectly rejected due to an off-by-one bounds check. (321041@main) (187306821)
  • Fixed an issue where black video-range pixel buffers were produced with super-black values instead of legal black. (321040@main) (187309125)

Networking

Resolved Issues

  • Fixed an issue where custom scheme CORS checks incorrectly blocked subresources loaded from an HTML document opened via a file: URL. (320685@main) (179999480)
  • Fixed an issue where reading a file-backed Blob range larger than 2GB truncated the read length due to an integer overflow. (320981@main) (187207663)
  • Fixed an issue where validation of an HTTP header value did not check its final character, allowing control characters and DEL to be accepted as valid. (321019@main) (187288970)

Performance

Resolved Issues

  • Fixed an issue where navigation could redundantly parse and decode a page’s URL multiple times, causing significant hangs for pages with very long URLs. (320292@main) (185796614)
  • Fixed excessive CPU and power usage caused by IntersectionObserver observation on pages with many observed elements. (320395@main) (185839711)
  • Fixed an issue where style resolution for elements sharing a scroll-timeline name could block the main thread for multiple seconds. (320902@main) (186090970)
  • Fixed an issue where ScrollingStateTree::insertNode performed redundant work reordering children on pages with many sibling scrolling nodes. (320527@main) (186143008)
  • Fixed an issue where ordinary property changes, such as toggling overflow: hidden, could trigger an unnecessary full-layer repaint. (320296@main) (186422878)

Rendering

Resolved Issues

  • Fixed an issue where a layer with a negative z-index inside a stacking context failed to composite correctly when overlapping an externally composited layer. (320473@main) (130595406)
  • Fixed an issue where the <mo> operator dictionary was consulted even when the element contained non-text content. (320730@main) (173078723)
  • Fixed an issue where accents and other combining marks could be painted outside their base letter when letter-spacing, word-spacing, expansion, or synthetic bold were applied. (320957@main) (186013305)
  • Fixed a regression that caused extra spacing to appear between an image and surrounding content near the bottom of a page. (320115@main) (186102748)
  • Fixed an issue where hovering links could trigger unnecessarily wide repaints, by repainting ink-overflow-only style changes without running inline layout. (320128@main) (186104270)
  • Fixed an issue where text nodes inside a display: contents element received redundant updates during style resolution. (320144@main) (186246705)
  • Fixed an issue where content changing inside an element with a blur() filter did not properly repaint, leaving stale pixels around the edge of the filtered region. (320398@main) (186278957)
  • Fixed an issue where a MathML rowspan="0" attribute was incorrectly clamped to 1. (320324@main) (186420659)
  • Fixed an issue where non-MathML children of <annotation> and <annotation-xml> were not rendered. (320334@main) (186429768)
  • Fixed an issue where drawing a native image did not respect the requested InterpolationQuality, affecting image scaling across canvas, SVG, and PDF rendering paths. (320429@main) (186439811)
  • Fixed an issue where a layer with clip-path or mask could be treated as fully opaque, causing incorrect rendering. (320634@main) (186722177)
  • Fixed an issue where a position: sticky element with -webkit-box-reflect applied could composite its reflection incorrectly. (320712@main) (186812790)
  • Fixed an issue where an optimization to text matrix handling when drawing glyphs could cause text to render upside down. (321067.5@safari-7626.1.8-branch) (187588379)

SVG

New Features

  • Added support for the SVG 2 path attribute on <textPath>, which lets text follow an inline path without referencing a separate <path> element via href. (321038@main) (185831618)

Resolved Issues

  • Fixed an issue where SVG conditional processing attributes on clipPath, mask, and pattern elements incorrectly prevented the resource from being referenced. (320207@main) (19103270)
  • Fixed an issue where SVG number attributes did not revert to their initial value when removed or set to an invalid value. (320226@main) (165378154)
  • Fixed an issue where a SMIL animation of a presentation attribute was never released after the animated element was detached from the document, permanently freezing the animated value. (320199@main) (185452156)
  • Fixed an issue where SVGLength resolved ex units by rounding the font’s x-height up to a whole pixel, causing the SVG DOM and getComputedStyle to disagree. (320606@main) (185984854)
  • Fixed an issue where additive color animations in SVG lost precision due to per-layer 8-bit rounding. (320166@main) (186223510)
  • Fixed an issue where an SVG transform change could leave an ancestor layer’s cached repaint rects stale, resulting in incorrect repainting. (320722@main) (187005898)
  • Fixed an issue where SVGImage::isAnimating() returned true for SVG images that had no animations. (320839@main) (187106396)

Security

Resolved Issues

  • Fixed an issue where a click in a backgrounded window could result in an action being taken. (320186@main) (186280350)
  • Fixed an issue where CredentialStorage::removeCredentialsWithOrigin failed to purge HTTPS credentials that used the default port. (320968@main) (187210150)

Web API

New Features

  • Added support for QuotaExceededError as a DOMException subclass, exposed on window and worker global scopes. (320856@main) (141586555)

Resolved Issues

  • Fixed an issue where scrollWidth and clientWidth returned inconsistent values when the page was zoomed in or out due to rounding. (320773@main) (121879134)
  • Fixed an issue where rotating the device to landscape orientation could pause a video playing in full screen, by adding a site-specific quirk. (320318@main) (148284059)
  • Fixed several URLPattern matching issues affecting Service Worker routing, including incorrect encoding of the username and password components and unhandled unclosed pattern groups. (320122@main) (183434701)
  • Fixed an issue where WebDriver reported an empty URL instead of about:blank for a browsing context that had not yet navigated. (320428@main) (183468933)
  • Fixed an issue where WKWebView.magnification reported the pre-gesture scale for a few moments after a pinch-to-zoom gesture ended, instead of the scale about to commit. (320341@main) (185381317)
  • Fixed an issue where Cache.delete() and Cache.keys() swallowed exceptions thrown while constructing a Request from a URL string. (320524@main) (186636105)
  • Fixed a regression where WebXR hit testing failed with a TypeError when offsetRay was not specified. (320717@main) (186997100)
  • Fixed an issue where interrupting a scroll with a tap could incorrectly trigger navigation of the link under the pointer. (321053@main) (187268555)
  • Fixed an issue where click events did not include the modifier keys that were pressed at the time of the click. (320984@main) (187270584)

Web Extensions

New Features

  • Added support for the offscreen API. (318833@main) (184269255)

Resolved Issues

  • Fixed an issue where i18n.getMessage() dropped non-string substitution arguments instead of coercing them to strings. (320931@main) (149855718)
  • Fixed an issue where webRequest.onBeforeRequest reported the resource type main_frame instead of sub_frame for subframe document loads. (321067@main) (167403388)
  • Fixed an issue where a high priority static declarativeNetRequest redirect rule was not favored over a low priority session redirect rule. (320739@main) (176176246)
  • Fixed an issue where webRequest events did not fire for loads blocked by a declarativeNetRequest rule. (320336@main) (176177851)
  • Fixed an issue where webRequest events did not fire for page loads created with tabs.create. (320430@main) (177727052)
  • Fixed an issue where an extension’s service worker lost its clients after being unloaded and reloaded, preventing it from finding an existing offscreen document. (320198@main) (186116488)
  • Fixed an issue where an extension’s popover could be cut off because vh, svh, and lvh units resolved to the popover’s minimum height instead of its actual size. (320726@main) (186391990)

Web Inspector

New Features

  • Added support for network throttling of the inspected page. (320893@main) (144892477)

Resolved Issues

  • Fixed an issue in Web Inspector where WebSocket handshake headers were missing when using the Network.framework loader. (320156@main) (185255819)
  • Fixed an issue in Web Inspector where a Debugger.globalObjectCleared event from a subframe cleared every target’s script map instead of only the affected target’s. (320787@main) (187042936)

WebAssembly

Resolved Issues

  • Fixed an issue where JS string builtins and imported string constants were not applied when compiling WebAssembly modules loaded as ES modules. (320174@main) (186037843)
  • Fixed an issue where the WebAssembly.Exception constructor rejected null options and incorrectly accepted non-object options. (320215@main) (186319813)
  • Fixed an issue where the WebAssembly validator rejected i31.get_s and i31.get_u when the operand’s static type was a bottom reference type. (320578@main) (186660590)

WebDriver

Resolved Issues

  • Fixed an issue where WebDriver BiDi’s script.call_function trimmed error messages incorrectly, cutting off part of the message. (320205@main) (186311752)
  • Fixed an issue where WebDriver’s deleteAllCookies command could report success before cookies had actually finished being deleted. (320440@main) (186410147)
  • Fixed an issue where WebDriver reported a stale element reference error instead of a no such element error for a node reference belonging to a different browsing context. (320518@main) (186615759)
  • Fixed an issue where WebDriver did not properly handle the right-hand variants of modifier keys, such as ShiftRight and ControlRight, on Cocoa platforms. (320741@main) (186961436)
  • Fixed a regression where WebDriver returned no such element instead of stale element reference for a node reference from a previous document. (320729@main) (187042183)
  • Fixed an issue where WebDriver’s Add Cookie command dropped the cookie when the request did not include a sameSite attribute. (320916@main) (187193380)

WebGL

Resolved Issues

  • Fixed an issue where WebGL contexts created with premultipliedAlpha: false were composited incorrectly. (320650@main) (53509398)

WebGPU

Resolved Issues

  • Fixed an issue where uploading a canvas as a WebGPU texture had very poor performance. (320765@main) (153642354)
  • Fixed an issue where GPUDevice.createBindGroup()‘s fast path failed to reuse bind groups containing a GPUBuffer entry alongside an external texture, forcing an unnecessary full rebuild. (320309@main) (186379208)
  • Fixed an issue where a WebGPU external texture imported from a video element using WebM or MSE playback rendered as a black placeholder instead of the video frame. (320427@main) (186470526)
  • Fixed an issue where using WebGPU render bundles on some GPUs could cause the device to be unexpectedly lost. (320689@main) (186688781)
  • Fixed an issue where WebGPU could report the primitive-index feature as available on GPUs that don’t support it, resulting in an uncaptured error. (320691@main) (186690340)
  • Fixed an issue in WebGPU where an array value constructor referenced through a type alias could generate invalid Metal Shading Language, silently producing incorrect compute results. (320690@main) (186803296)
  • Fixed an issue in WebGPU where inlining three or more calls to a function operating on a large struct could silently miscompile on some devices, producing incorrect results. (320697@main) (186804865)
  • Fixed an issue where asynchronous pipeline creation could stall rendering. (321042@main) (187296614)

WebRTC

Resolved Issues

  • Fixed an issue where a WebRTC UDP connection to an unresolved remote address was incorrectly restricted to IPv6. (320716@main) (186844205)
  • Fixed an issue where a TURN UDP connection could fail unnecessarily when DNS resolution failed, by allowing the socket to open directly to the host. (320720@main) (186847964)

September 23, 2026 09:46 PM

September 21, 2026

Igalia WebKit Team: WebKit Igalia Periodical #78

Igalia WebKit

Update on what happened in WebKit in the week from September 14 to September 21.

Following up to last week, we have a handful of exciting fixes to WebKitGTK and WPE WebKit covering Sysprof, rendering of CSS "filter" and "backdrop-filter", and the Layer-Based SVG Engine. We also have optimizations, stable releases, and new blog posts!

Cross-Port 🐱

Fixed the pairing of begin and end marks in SysprofAnnotator, which kept only one begin per mark and therefore mistimed recursive, concurrent and overlapping marks, so Sysprof traces of WebKitGTK and WPE WebKit now show correct durations for nested work.

Graphics 🖼️

Fixed CSS filter and backdrop-filter rendering on transformed elements in the Skia compositor used by WebKitGTK and WPE WebKit, where drop-shadow always fell in the same direction on screen instead of turning with the element, blur was scaled by neither the transform nor the device scale factor, and perspective was not handled at all. The filtered subtree is now painted with the layer transform on the canvas into a surface in the layer's own plane at a fixed scale, with the filter applied there and the result drawn back through the layer transform, as Chromium does.

Stopped rasterizing the parts of a tile that the clip discards in the Skia backing store used by the GTK and WPE ports. Both the drawn quad and its source rectangle are now cut down to the clip first, so the rasterizer no longer fills pixels that are immediately thrown away when a tile is larger than the clip.

Fixed the transform attribute on the outermost <svg> element, which the legacy SVG engine ignored entirely and the Layer-Based SVG Engine (LBSE) wrongly applied inside viewBox space. The attribute is now turned into the CSS transform property, one CSS function per SVG function, so that it transforms the element's CSS box, matching the behaviour of Chromium and Firefox.

Releases 📦️

After another six-month development cycle, WebKitGTK 2.54.0 and WPE WebKit 2.54.0 have been published. These are the first stable releases of the 2.54 series, that as a highlight include a completely new compositor based on Skia. As it is becoming tradition with each major release, two blog posts go into details of what's new in WPE WebKit 2.54, and in WebKitGTK 2.54.

This release is a milestone for WPE WebKit: after a couple of years in the incubator, the new WPEPlatform embedding API is now considered stable and recommended for new developments. As part of the stabilization, the WPEPlatform manual now includes not one, but two introductory tutorials, and a migration guide to aid migration from the (now) legacy libwpe-based embedding API.

Note that the libwpe-based legacy API is still available and will be kept in light maintenance mode for the foreseeable future.

Community & Events 🤝

Coinciding with the first stable releases, Carlos García has written about the new Skia compositor that is debuting in the 2.54 series. It includes plenty of details on why and how the new compositor has been built to replace the aging TextureMapper, and measurements that demonstrate how it is bringing graphics performance improvements across the board.

At the same time, Nikolas Zimmermann published a companion article about damage handling in the new Skia compositor that explains how the feature it has been rewritten with a focus on correctness while ensuring that it saves doing unnecessary work without incurring on excessive busywork that would cancel the benefits.

That’s all for this week!

By Igalia WebKit Team at September 21, 2026 08:15 PM

Carlos García Campos: Skia compositor for WPE WebKit and WebKitGTK

Igalia WebKit

WPE WebKit and WebKitGTK 2.54 have been released with a bunch of improvements and new APIs as usual, but there’s one point that kept the Igalia WebKit graphics team busy for the whole cycle: the new Skia-based compositor. The replacement of Cairo with Skia for content rendering has been a success and it’s already well integrated and optimized. We thought we could try to use Skia for the composition too and replace TextureMapper with Skia. TextureMapper was introduced in 2010 for the Qt port and later adopted by other ports. It uses the OpenGL ES API and maintains a collection of shader programs to paint different content. Nowadays TextureMapper is mostly the same code and shader programs, and it’s unmaintained and missing features. However, the performance was good and it has served us really well all these years. So, this time the goal was not to get better results in benchmarks, but to modernize the implementation, reduce the amount of code to maintain ourselves (like all shader programs) and make it easier to implement the missing features and fix existing bugs. This post is a summary of all the work we have done this cycle to implement the new Skia compositor.

SkiaCompositingLayer

The first step was adding an SkiaCompositingLayer class to replace TextureMapperLayer and adapt all the code to use one or the other depending on an environment variable. The initial implementation was based on the TextureMapper one for the things that are common like iterating the layer tree, computing transformations, etc. The way layers produced their contents didn’t change, so we were receiving textures for tiled content, video buffers, WebGL, accelerated 2D canvas, etc. SkiaCompositingLayer created a Ganesh Skia surface to draw those textures using SkCanvas::drawImageRect(). This initial implementation was enough to run the default MotionMark test suite, since it doesn’t use other composition features. Even though performance was not the goal, we had to make sure we didn’t regress. This initial implementation was neutral in MotionMark. We needed tests to implement those features and measure performance at the same time, so we decided to add a new set of tests to MotionMark, just extending the existing tests to require composition, which makes sure that filters, masks, path clipping, transformations, etc. were done by the compositor.

Filters

We first tried implementing filters using an intermediate surface like TextureMapper does. It worked, but the MotionMark score in the filters test was much worse. We realized that with Skia we could implement most of the filters without using an intermediate surface. All filter types except blur and drop shadow can be simplified to an SkColorFilter with SkImageFilter::asAColorFilter() which can be implemented without an intermediate surface, just by setting the color filter in the SkPaint we pass to SkCanvas::drawImageRect(). This not only fixed the performance regression, but also gave better results than TextureMapper, which always needs an intermediate surface.

Masks

There are two different kinds of masks: image mask, where the source mask is an image already, and clip path, where the mask is represented by a path to be clipped. In TextureMapper both are implemented the same way using intermediate surfaces. The mask is painted into a surface and then the masked layer creates an intermediate surface where its contents are first painted and then the mask contents on top using DstIn blend mode. Skia has APIs that allowed us to implement both cases in a much simpler and more efficient way. In the case of image masks, where we already have an image, we paint the mask contents once and keep it cached, and then the masked layer creates an SkShader for the image mask that is passed to SkCanvas::clipShader() without having to paint into an intermediate surface. Clip path masks are even easier, because we can just take the path we get and build an SkPath we can pass to SkCanvas::clipPath(), without having to paint the mask as an image at all or use any other intermediate surface. Once again, masks were not only easier to implement but they ended up being more performant too.

Grouped bar chart of ten MotionMark composition subtests, comparing TextureMapper with the Skia compositor. Filters, clipping and mask tests are three to four times faster with Skia; the three leaves tests are about 20% slower.
MotionMark composition suite, WPE with GPU rendering on a Raspberry Pi 4, comparing TextureMapper (312400@main) with the Skia compositor (313600@main). TextureMapper never implemented blend modes, so its high score on bouncing blend circles is the score for not doing the work.

3D contexts

The implementation of 3D layer contexts is fairly independent of TextureMapper and OpenGL, so we could just take it almost as it was, using SkPath to build the clips and a few other adaptations. We could also fix existing bugs like the z-ordering that has always been broken in TextureMapper.

Two screenshots side by side of the same page. Under TextureMapper a small red box sits flat on top of a green plane rotated in 3D. Under the Skia compositor the red box is much taller and is cut by the plane: a sliver shows past the left edge, the middle is hidden behind the plane, and the right part is drawn in front of it.
The same page rendered by TextureMapper (left) and by the Skia compositor (right), WPE on the same build. The red box intersects the rotated green plane. TextureMapper draws the box flat against the plane, so the intersection is lost; the Skia compositor splits it, drawing the part in front of the plane and hiding the part behind it.

Blend modes

TextureMapper never supported blend modes and they were easy to implement with Skia just using the SkPaint property for it. This made several layout tests start passing.

Batched painting

After implementing all the features we were at a point in which we had the same or better performance in all tests except for three MotionMark compositing tests that were giving much worse results. Those tests use small layers and give a high result which means we end up adding a lot of layers to the scene before we start skipping frames. The root cause was the large number of layers filling the command queue of Ganesh. Skia Ganesh queues the GL drawing operations instead of sending them to the GPU right away. When the surface is flushed for whatever reason, the queued GL drawing operations are then processed and sent to the GPU. This allows Skia to apply nice optimizations like merging several tasks and reducing the amount of draw operations we end up sending to the GPU. In those tests where a lot of layers are created and painted to the compositor Skia surface the internal command queue ends up being huge too. Processing and analyzing such a long queue to optimize what we send to the GPU required more CPU work than what we save by optimizing the GL draw operations. Skia provides an API that allows us to do the batching ourselves. Since the compositor already has information to decide what operations could be merged together, we could reduce the internal queue size in many cases. We can merge SkCanvas::drawImageRect() operations as long as they share the same color filter, blend modes and sampling options. In the best case scenario we could reduce the whole internal queue to just one operation. This time the change improved the results of those tests getting them to about 93% of the TextureMapper score, but still a bit behind.

Promise images

The Skia Ganesh backend requires that an SkImage backed by a texture is created for the current thread GrDirectContext, even if it’s borrowing an existing texture. In WebKit all textures are created with a sharing GL context so that they can be accessed and destroyed from different threads with the same sharing GL context. So, for a layer whose content is an image we had to create a texture in the compositing thread to upload the pixels if the image was not accelerated, or for accelerated images get the texture identifier of the image, and then create another SkImage from the compositing thread borrowing the texture for the current GrDirectContext. The Skia Ganesh backend provides an API to create promise images, which can be created from any thread but targeting a specific thread, providing a fulfill callback that will be called on the target thread when the SkImage is first used to retrieve the wrapped texture. This way we can create the SkImage from the main thread for the compositing thread without using OpenGL at creation time. For non-accelerated images we realized we don’t need to manually create the texture and upload the pixels in the compositor, we can just pass the unaccelerated SkImage to the compositor SkCanvas and Skia will handle it internally much more efficiently than we did. And this change improved those compositing tests much further than we expected. The reason turned out to be the batching from the previous section: Skia merges the entries of an image set by comparing texture proxy pointers, and until now we were wrapping the texture in a new SkImage on every frame for every layer, so hundreds of layers drawing the very same image produced hundreds of different proxies that Skia could not merge. Passing the same SkImage every time collapses all of them into a single draw operation, which is the best case we described above. With batched painting and promise images together we could beat TextureMapper significantly.

Line chart of the three MotionMark leaves subtests by WebKit revision. All three step up sharply at revision 314626 when batched painting landed, and again at revision 315529 when promise images landed.
MotionMark composition suite, leaves subtests, WPE with GPU rendering. Score per revision; higher is better. The same two steps appear with CPU rendering.

Deferred Display Lists (DDL)

When we switched to Skia for painting, we kept the threaded rendering model, just using a separate smaller queue for GPU rendering workers. The GPU workers created their own GrDirectContext to paint the layer tiles. The resulting textures were re-wrapped in the compositing thread for the compositor GrDirectContext using fences for the proper synchronization. We knew this was not the recommended way to use Skia Ganesh from multiple threads, but with TextureMapper we had no other option. However, with the Skia compositor we can do it the recommended way by using a single GrDirectContext in the compositing thread and use Deferred Display Lists (DDL) and promise images to paint the tiles. With DDL, GPU workers no longer use GL at all and they don’t need a GrDirectContext, they paint tiles into a display list that records the GL drawing operations, but without touching GL. For image drawing operations recorded into the DDL, promise images are used too. Since this is now all CPU work we can remove the smaller GPU worker queue and use a single queue with more workers. The compositor replays the DDL into an SkSurface that is then passed to the compositor SkCanvas.

This change fixed rendering glitches on Android and was performance neutral for the whole composition suite and for most of the MotionMark tests, but in MotionMark 1.3 at 15fps it cost 29% in Suits and 14% in Leaves, while improving Images by 9%. Correctness and the other benefits of DDL made us accept those regressions.

Line chart of the MotionMark suits score by WebKit revision. The score drops from about 470 to about 335 when deferred display lists are enabled, returns to 470 when they are disabled, and drops again when they are re-enabled, staying there.
MotionMark 1.3 at 15fps, suits subtest, WPE with GPU rendering on a Raspberry Pi 4. Shaded regions are where deferred display lists were enabled by default. CPU rendering moves less than 1% at all three switches, since it has no GPU worker threads for DDL to change.

Damage

TextureMapper already supported using damage information to optimize the painting while compositing, but it has always been disabled at run time because there were issues we never managed to fix. With the Skia compositor we decided to start from scratch and properly handle the damage information while compositing to render only the parts of the frame that actually changed. I’m not going to go into detail here because Nikolas Zimmermann has written an amazing blog post about it with all the details.

Current situation

The Skia compositor is finished and enabled by default in 2.54. Even though it was not the main goal, it performs better than TextureMapper in most of the benchmarks we run: the composition suite we added is 45% faster, and MotionMark 1.3.1 is 35% faster. The exception is MotionMark 1.3 at 15fps with GPU rendering, which comes out flat, because the Suits and Leaves tests are still about 26% and 10% behind due to the deferred display lists trade-off described above.

We are already working on fixing existing issues in composition that we never fixed in TextureMapper. In the main branch TextureMapper is now disabled by default at build time, and support will be removed soon for the GTK and WPE ports. In 2.54 it’s still a run-time decision so if you find any issue with 2.54, you can check if it’s a Skia compositor regression by trying TextureMapper with WEBKIT_USE_SKIA_FOR_COMPOSITION=0 environment variable.

Bar chart of overall benchmark scores today relative to the TextureMapper baseline. The composition suite is 45% ahead with GPU rendering, MotionMark 1.3.1 is 35% ahead, and MotionMark 1.3 at 15fps is level with GPU rendering and 10% ahead with CPU rendering.
Overall (geometric mean) score, WPE on a Raspberry Pi 4: 320000@main and later against the TextureMapper baseline at 312400-313296@main. Bars start at the baseline. Part of the gain in the MotionMark suites is Skia rendering work rather than the compositor.
Bar chart of MotionMark 1.3 at 15fps subtests, showing the change from the TextureMapper baseline to today. Suits is 26% behind and leaves 10% behind with GPU rendering, while both are well ahead with CPU rendering; every other subtest is level or ahead.
WPE on a Raspberry Pi 4, change from the TextureMapper baseline (312400-313296@main) to 320000@main and later. Suits and leaves are the deferred display lists trade-off, not the compositor switch, which was neutral in this suite.

Future plans

We are already working on further improvements like using promise images for all external textures we have to pass to the compositor. We will explore the possibility of using Vulkan with the Ganesh backend instead of GL and eventually try the new Graphite backend. And of course we will continue fixing any existing issues related to the compositor.

By carlos garcia campos at September 21, 2026 08:40 AM

Nikolas Zimmermann: Bringing damage to the Skia compositor

Igalia WebKit

Teaching the new Skia compositor to recomposite only what changed

September 21, 2026 12:00 AM

September 16, 2026

WPE WebKit Blog: WPE WebKit 2.54 highlights

Igalia WebKit

The WebKit team at Igalia is happy to announce a new release series of WPE WebKit. This release has two main highlights: the new WPEPlatform API, now stable and enabled by default, and a web process compositor built on the Skia graphics library, replacing the TextureMapper-based one. Read on for the details on both, along with a summary of the other most noteworthy changes from the latest release cycle.

WPEPlatform: the new WPE API

WPEPlatform, the API that has been in the works for the past several release cycles, takes center stage in 2.54: it is now enabled by default and its API is considered stable, so applications can build on it without expecting breaking changes in future releases. Consequently, the traditional libwpe-based API is officially deprecated: it remains available and maintained, but new code should target WPEPlatform, and existing embedders are encouraged to plan their migration. This also applies to Cog, which will not have stable releases beyond the 0.18.x series.

The best part of the new API is how much simpler it is. Under libwpe, an application had to create a view backend (usually through WPEBackend-fdo’s “exportable” backend) and drive rendering, buffer management, and input dispatch itself through its callbacks. WPEPlatform moves all of that into WebKit and the platform implementation, so migrating an application is mostly a matter of deleting code: in the common case, the application constructs its WebKitWebView without a backend, WebKit selects a suitable platform automatically, and everything built on top of the web view carries over unchanged. The platform API only surfaces when an application wants more than the defaults: pinning to a particular platform, handling raw input events, or driving the toplevel window. Each of those is a few lines against a small GObject API rather than a set of C callbacks to implement.

To help embedders make the move, WPEPlatform is now extensively documented. The reference documentation includes an overview of the platform API (covering its relationship to libwpe and what is intentionally not part of the new API), a guide on compiling against it, a tutorial on writing a browser, and a tutorial on writing a WPE platform implementation. A guide on migrating from libwpe, with a symbol-by-symbol mapping table, walks embedders through the process step by step.

Since WPEPlatform is now built by default, the wpe-platform-2.0 pkg-config module is available in regular builds. Conversely, the legacy API can be disabled at build time with ENABLE_WPE_LEGACY_API=OFF when it is not needed.

Additions and final adjustments

New platform APIs added this cycle include:

  • WPEProcessManager and WPEProcessLaunchOptions, which allow the embedder to control how the auxiliary WebKit processes are launched and terminated. This is particularly important on Android, where each process is a service that must be started with bindService(), and it removes the last reason a WPEPlatform-only build could not work there. Note that, unlike the rest of WPEPlatform, the process management API is only built when targeting Android and remains experimental: it is not yet generic enough to be enabled everywhere, and it may still change in future releases.
  • Gamepad rumble support, through wpe_gamepad_has_rumble() and wpe_gamepad_rumble(), with a built-in implementation based on libmanette. This enables the Gamepad API vibrationActuator for web content.
  • A new WPE_SETTING_OVERLAY_SCROLLBARS setting, enabled by default, which may be disabled by applications to opt into classic, always-visible scrollbars.
  • A new WPE_INPUT_PURPOSE_SEARCH input purpose, allowing input methods to detect when the values of an input field are expected to be search terms.

A few final adjustments were made to the API before declaring it stable, which may require updates to platform implementations and applications developed against the earlier previews:

  • The WPE_SETTING_DISABLE_ANIMATIONS setting has been replaced by WPE_SETTING_REDUCED_MOTION, matching the dedicated reduced-motion setting introduced in GNOME 50, and a new tri-state WPE_SETTING_INTERFACE_CONTRAST setting has been added. Together with the existing WPE_SETTING_DARK_MODE setting, these make the prefers-reduced-motion, prefers-contrast, and prefers-color-scheme media queries follow the platform settings.
  • wpe_gesture_controller_handle_event() now returns a boolean indicating whether the event was consumed.
  • The WPE_DMABUF_BUFFER_FORMAT environment variable has been renamed to WPE_BUFFER_FORMAT.

Beyond Linux: Android

A good measure of the new API’s maturity is wpe-android, which has been rebuilt this year as a WPEPlatform platform implementation living entirely outside the WebKit tree: Android’s display system now looks to the engine like any other WPE platform, and applications get a convenience Java API modeled after android.webkit.WebView. The new WPEProcessManager API removed the last dependency on libwpe there. You can read more about it in this post.

On the WebKit API side

The shared WebKit API has also seen additions this cycle:

Graphics improvements

A new Skia-based compositor

This cycle brings the largest overhaul of the rendering architecture since the adoption of Skia for 2D rendering: the web process compositor now uses the Skia API instead of the venerable TextureMapper. Layers are composed into the final frame using Skia, which allows sharing a single rendering infrastructure across the whole graphics stack and enables several optimizations:

  • Tile contents are recorded into deferred display lists and replayed on the compositor thread, so painting worker threads no longer need to touch the GPU at all.
  • Batched painting groups the drawing of many layers into a single Skia call, which improves performance on pages with many layers that can be painted in the same operation.
  • Unnecessary clip operations are avoided whenever possible, keeping the batched paths effective.

Beyond raw performance, expressing compositing as Skia draw calls made several features simpler and faster: filters and masks no longer require intermediate offscreen surfaces in most cases, and CSS blend modes, which TextureMapper never implemented, now work in composited layers. And since Skia can target both OpenGL and Vulkan, the compositor no longer stands in the way of Vulkan-based rendering in the future.

The new compositor also works when using the legacy libwpe-based API.

The consolidation on Skia goes beyond compositing: the option to use Cairo for 2D rendering has been removed, making Skia the only 2D rendering implementation. Rendering tiles in the main thread is no longer supported either, so threaded rendering is now the only tile painting path.

Damage-aware compositing

Damage tracking has seen substantial work this cycle. The damage is the region of the view that changed since the previous frame and therefore requires repainting; Paweł Lampe’s introduction to damage propagation covers the concept in depth. Compositing itself now uses this information: each draw the compositor issues is restricted to the damaged rectangles, in a way that preserves the batched painting described above, and damage propagation to the platform is now enabled by default. A new DamageRectangleThreshold preference allows embedders to tune the balance between damage precision and bookkeeping cost.

The performance impact

What drove the compositor rewrite was making it simpler and easier to maintain, as Carlos García Campos explained at the Web Engines Hackfest 2026; performance came later, since TextureMapper started out ahead after more than a decade of tuning. By now the optimization work has more than closed that gap. The public WPE performance dashboard, which continuously runs benchmarks on Raspberry Pi 4 devices, tells the story: comparing the last TextureMapper-based revisions against the current ones, and thus measuring the cumulative effect of the graphics work described in this section, the MotionMark score is up by around 36%. On the composition-focused variant of the benchmark, where the compositor dominates the workload, the score is up by around 45%.

Bar chart comparing MotionMark scores on a Raspberry Pi 4: MotionMark 1.3.1 at 30 FPS goes from 246 with TextureMapper to 334 with the Skia compositor (+36%), and the Composition variant from 131 to 189 (+45%)

The dashboard also records the GPU load during each run, and there the difference is even more telling. MotionMark increases scene complexity until the browser can no longer sustain the target frame rate, so a higher score already means more work per frame; the Skia compositor delivers that while keeping the GPU markedly less busy. On embedded devices, where the GPU is typically the scarcest resource, this headroom translates directly into smoother pages.

Bar chart comparing GPU load during the same benchmark runs: 56% with TextureMapper against 39% with the Skia compositor on MotionMark 1.3.1, and 47% against 26% on the Composition variant

One part of the work goes largely unmeasured here, though. MotionMark animates nearly the entire viewport, so there is hardly anything for damage tracking to save. Real-world content behaves differently: usually a small area of the page is changing while everything else stays still, and skipping all of that quiet area cuts the per-frame GPU work to a fraction.

Other rendering improvements

A GPU atlas is now used for batched raster image uploads, regardless of the compositor in use, and it is reused across frames when the image set does not change, avoiding needless texture allocation and pixel uploads.

Asynchronous scrolling is smoother: several synchronization issues between the main, scrolling, and compositing threads that caused glitches while scrolling have been fixed, and the scrolling thread no longer blocks the compositor to flush its state, removing input-latency stalls on pages with many layers.

Finally, more animations can now run on the compositing thread: CSS animations using the steps() and linear() timing functions no longer force main-thread animation.

Better multimedia on embedded hardware, and a WebRTC transition

Let’s start with the transition: the GStreamer-based WebRTC backend is being replaced with a LibWebRTC-based implementation, which is expected to be available in the next release cycle. As a consequence, WebRTC support, which in previous releases required building with experimental features enabled, is disabled in 2.54.

The rest of the multimedia work moved forward at full speed, with a strong focus on embedded hardware:

  • Hardware video decoding and encoding for platforms with a Qualcomm GPU has been added, leveraging the qtic2vdec and qtic2venc GStreamer elements.
  • Video decoding limits are now respected in MediaCapabilities queries, and can be overridden with the WEBKIT_GST_VIDEO_DECODING_LIMIT environment variable, so a single build can serve devices with different capabilities.
  • Resource usage on pages containing many videos has been improved, by stopping the pipelines of muted, invisible video elements.
  • A new feature flag, enabled by default, allows low-end devices to skip caching pages with multimedia content in the back-forward cache, so that a suspended pipeline cannot hold a scarce hardware decoder hostage.
  • Media capability reporting is more accurate: non-AAC mp4a codecs (MP3, AC-3, E-AC-3) are correctly reported as supported when decoders are present, xHE-AAC support is auto-detected, and Dolby AC-4 is advertised for MSE on systems that support it.
  • Experimental support for SourceBuffer.changeType() has been added to the MSE backend, along with a fix for playback stalling at ad transitions on Twitch.
  • Persistent licenses are now supported in the Thunder CDM for encrypted media.

On top of this, the GStreamer backend has received a substantial amount of memory-safety and lifetime-correctness work that translates into a more stable multimedia experience.

WebXR

The OpenXR-based WebXR implementation continues to progress. The main highlight is support for the WebXR Layers API: quad, cylinder, equirect, and cube layers are now implemented, in addition to the already supported projection layers. Layers can be backed by texture arrays, and XRSession.maxRenderLayers lets content query the compositor’s layer budget.

The backend has also been decoupled from OpenGL ES through an abstract graphics binding, paving the way for a future Vulkan-based binding.

WebXR support remains a build-time option, enabled with the ENABLE_WEBXR=ON CMake option; the Layers support additionally requires ENABLE_WEBXR_LAYERS=ON.

Other improvements to the WPE port

Several long-standing gaps in the WPE port have been closed this cycle:

  • A built-in popup menu is now used as the default implementation for <select> elements when the WebKitWebView::show-option-menu signal is left unhandled, so option menus work out of the box.
  • Initial drag-and-drop support has been added: web views now handle drags driven by the mouse events they already receive.
  • Spell checking is now supported using the Enchant library, enabled by default, and can be toggled at build time with the ENABLE_SPELLCHECKING CMake option.
  • The on-screen keyboard is no longer shown when an element is focused programmatically; it only appears as a result of user interaction.
  • Initial support for the Pointer Lock API can be enabled at build time with the ENABLE_POINTER_LOCK CMake option.
  • Saving files from the Web Inspector now works in WPE.

Web Platform support

As usual, this list is not exhaustive as WebKit continuously progresses in its support for new standards. Some of the highlights for this release are:

What’s new for WebKit developers?

WebKit can now use mimalloc as its memory allocator, as an alternative to its own bmalloc. For now it is the default only on some architectures (32-bit ARM, MIPS, RISC-V, and builds supporting 64 KB memory pages); everywhere else bmalloc remains the default, and mimalloc can be enabled with the USE_MIMALLOC build option.

Logging now falls back to the standard error output when journald is not reachable, which is common in minimal containers, and the new WEBKIT_DEBUG_OUTPUT environment variable allows choosing the log destination explicitly.

Profile-guided optimization is now supported in regular CMake builds with Clang, through the ENABLE_LLVM_PROFILE_GENERATION and USE_PGO_PROFILE options.

Finally, a note for packagers: building WPE WebKit now requires Ninja, as the CMake Makefile generator is no longer supported.

Looking forward to 2.56

The 2.56 release series will bring even more improvements, and we expect it to be released during the spring of 2027. Until then!

September 16, 2026 12:00 AM

September 14, 2026

Igalia WebKit Team: WebKit Igalia Periodical #77

Igalia WebKit

Update on what happened in WebKit in the week from September 7 to September 14.

In this week's edition, the PNG image decoder received a much needed cleanup, as well as another swipe of unassorted graphics improvements, most notably in the Layer-Based SVG Engine. Finally, we also have an exciting and informative article about how to use the WPE Platform API on Raspberry Pi.

Cross-Port 🐱

Graphics 🖼️

Cleaned up the PNG image decoder a little bit, by removing conditional code that was used to support libpng versions older than 1.5.0, and requiring that as the minimum version. Given that version 1.5.0 was released back in 2011, it is expected that every system where current WebKit works will have much newer versions of libpng anyway.

Fixed an assertion failure for an outermost <svg> with non-visible overflow in the Layer-Based SVG Engine (LBSE) by refreshing its scroll dimensions at the end of layout, a first step towards reworking <svg> scrolling, which is not yet spec compliant.

Fixed an assertion failure when computing filter outsets for <feGaussianBlur> and <feDropShadow> with a negative stdDeviation, which turns the primitive off but was still passed on to the outset calculation, affecting both SVG filters and CSS reference filters.

Stopped reporting damage every frame for composited layers that use CSS filter in the Skia based compositor, so a blurred or drop-shadowed element is now only repainted when its subtree, or the applied (possibly animated) filter value, actually changed.

Fixed the bug that composited layers with clipping or masking and fully covered by opaque child layers were treated as opaque layers.

Fixed two debug assertion failures in the Layer-Based SVG Engine (LBSE), one where an SVG transform change left ancestor layer repaint rects stale and one where repainting on a compositing change during layout tripped an over-strict check on the paint offset cache.

Fixed the white noised image issue on the combination of i915 driver and Intel Arc after 308458@main introduced image atlas uploading.

Community & Events 🤝

Published a blog post describing the steps to try WPE Platform API on Raspberry Pi with latest WebKit main branch.

That’s all for this week!

By Igalia WebKit Team at September 14, 2026 07:07 PM

September 13, 2026

Hironori Fujii: Asynchronous scrolling for touch events in WPE and WebKitGTK

Igalia WebKit

The WPE and GTK ports have supported touch events for a long time, but asynchronous scrolling only worked for wheel events. Scrolling driven by touch still depended on the web process main thread. This change puts touch events on the asynchronous scrolling path too.

Let’s start with the background: what asynchronous scrolling is and why it needs to be built differently here.

What is asynchronous scrolling? #

Why it is needed #

A naive implementation of scrolling looks like this:

  1. The UI process receives an input event (wheel or touch).
  2. It sends the event to the web process main thread.
  3. The main thread runs the page’s JavaScript event listeners.
  4. If nothing called preventDefault(), the scroll position is updated.
  5. The page is rendered at the new scroll position.

Step 3 is the problem. The main thread is easily blocked for hundreds of milliseconds by JavaScript execution or layout, and scrolling is frozen for that whole time. Your finger moves, the screen does not. That is what synchronous scrolling feels like.

Asynchronous scrolling moves the work off the main thread: the scrolling thread updates the scroll position, and the compositor thread composites and presents the frame. Neither needs the main thread, so scrolling keeps running at 60fps even when the main thread is busy.

The scrolling tree and event regions #

Two data structures make this possible.

The scrolling tree is a tree of the scrollable areas of a page. The main frame, overflow: scroll elements, position: fixed/sticky elements and so on each become a node holding its own scroll position and a reference to its layer. The scrolling thread updates scroll positions by looking only at this tree, and the compositor thread then draws the layers at their new positions — the main thread is not involved in either step.

But there are places where scrolling on its own would be wrong, because the page might call preventDefault() from an addEventListener("touchstart", ...) handler. That is what event regions are for.

An event region records, per layer and at rendering time, “this rectangle has a listener for this kind of event”. EventRegion keeps separate regions for touchstart, touchmove, pointerdown, mousedown and friends, and for any given point it yields a TrackingType:

enum class TrackingType : uint8_t {
NotTracking = 0, // No listener. The event does not even need to be delivered.
Asynchronous = 1, // Passive listeners only. Scroll now, notify the page later.
Synchronous = 2 // A non-passive listener may call preventDefault(). We must wait.
};

The passive distinction is what makes Asynchronous possible. preventDefault() cancels an event only if the listener was registered with passive: false; from a passive listener it does nothing. And on window, document and document.body, touchstart/touchmove (and wheel) default to passive: true — see MDN: Using passive listeners. So Asynchronous is the case “there are listeners, but none of them can cancel the scroll”: start scrolling now, deliver the event to the main thread afterwards.

Because this information travels to the scrolling thread along with the layer tree, an incoming input event can be classified without waking the main thread. That is the heart of asynchronous scrolling.

Why the iOS implementation could not be reused #

The iOS port already implements asynchronous scrolling for touch events, but it could not be reused, because the process layout is different.

iOS:
  UI process:  platform layer tree + scrolling tree + touch event input
  Web process: main thread (DOM, layout)

WPE / GTK (Coordinated Graphics):
  UI process:  touch event input only
  Web process: main thread (DOM, layout)
               + EventDispatcher thread / scrolling thread
               + platform layer tree + scrolling tree

On iOS both the platform layer tree and the scrolling tree live in the UI process — the very process that receives touch events — so the classification and the scroll both happen right there. On WPE and GTK we use Coordinated Graphics, and both trees live in the web process instead. The classification therefore has to happen after sending the event to the web process, but before touching the main thread.

Fortunately the same problem was already solved for wheel events. The web process has an EventDispatcher thread that receives wheel events from the UI process without going through the main thread and consults the scrolling tree directly. This change builds the same shape for touch events.

How a touch becomes a scroll in WPE #

One more piece of background: in WPE a touch does not scroll the page directly. Touch events are first offered to the page; only if the page does not consume them does the UI process turn the touch sequence into scrolling.

The decision point is PageClientImpl::doneWithTouchEvent(). If the page handled the event, gesture detection is cancelled with wpe_gesture_controller_cancel() so the engine does not also act on it. If it was not handled, the event is fed to the WPE platform gesture controller via ViewPlatform::handleGesture(), and a recognized WPE_GESTURE_DRAG is turned into a synthetic scroll event pushed back into the page as a wheel event:

GRefPtr<WPEEvent> simulatedScrollEvent = adoptGRef(wpe_event_scroll_new(
m_wpeView.get(), WPE_INPUT_SOURCE_TOUCHSCREEN, 0, static_cast<WPEModifiers>(0), dx, dy, TRUE, FALSE, x, y));
page().handleNativeWheelEvent(WebKit::NativeWebWheelEvent::create(simulatedScrollEvent.get(), phase));

That TRUE is precise_deltas: touch-driven scrolling in WPE reaches the engine as precise-delta wheel events, which becomes relevant later.

The important consequence is this: the UI process cannot start scrolling until it knows whether the page is going to consume the touch. That answer used to come from the web process main thread — so when the main thread was busy, scrolling did not start. That is the problem this change fixes.

The change #

Enabling touch event regions #

A new ENABLE(COORDINATED_TOUCH_EVENTS) is introduced in PlatformEnableGlib.h, and it turns on ENABLE(TOUCH_EVENT_REGIONS) whenever touch events are enabled on WPE/GTK. The AlwaysUseTouchEventRegions preference now defaults to true under that flag, so Document::shouldUseTouchEventRegions() returns true and touch regions are actually recorded on the layers during rendering.

UI process: send to the EventDispatcher instead of the main thread #

The old WebPageProxy::handleTouchEvent() consulted a touchEventTracking state kept in the UI process and sent Messages::WebPage::TouchEvent, i.e. straight to the web process main thread.

The new version delegates all classification to the web process and only queues events and delivers answers. One event is in flight at a time; the next is sent when the reply arrives. The flood of touchmove events produced while a finger moves is coalesced into the newest queued event when that is also a touchmove, and the coalesced events are flushed to doneWithTouchEvent() together with the reply.

The destination is now Messages::EventDispatcher::TouchEvent.

Web process: classification on the EventDispatcher thread #

EventDispatcher::touchEvent() runs on the EventDispatcher thread, where it looks up the page’s scrolling tree, asks it for a TrackingType, and splits three ways:

  • NotTracking — no listeners. Reply handled = false immediately, without bothering the main thread at all. The UI process can start scrolling right away.
  • Asynchronous — passive listeners only, so nothing can cancel the event. Reply handled = false first so scrolling starts, then deliver the event to the main thread.
  • Synchronous — a non-passive listener may call preventDefault(), so wait for the main thread result as before.

Replying without a main thread round trip is possible because the new TouchEvent message is declared AnyThread in EventDispatcher.messages.in. The iOS equivalent is MainThreadCallback, which always replies from the main thread.

If there is no scrolling tree for the page yet, the event goes to the main thread as before.

Classifying a touch in the scrolling tree #

The classification itself is ScrollingTreeCoordinated::eventTrackingTypeForTouchEvent(). It works in two stages.

First, for each newly pressed touch point: convert the point from view to contents coordinates, hit test the layer tree down from the root contents layer, take the frontmost layer whose event region contains the point, and query that region. It is queried for many event types, because a touch fires more than the DOM touch* events — pointer*, compatibility mouse events and gesture* too, and a non-passive listener for any of them forces synchronous handling. The results are folded into a small TouchEventTracking struct with four fields: start, move, end and force-change.

Second, the tracking type of the event as a whole is derived from the touch point states, merging the per-field values. Merging picks the stronger of two types (NotTracking < Asynchronous < Synchronous), so if any single point needs synchronous handling, the whole event is synchronous.

TouchEventTracking persists for the lifetime of a touch sequence and is reset once all points are released, so the hit test done at touchstart is reused for the following touchmove/touchend. That guarantees a sequence never flips from synchronous to asynchronous halfway through just because a finger moved off a listener’s area.

<input type=range> #

A slider handles touches internally even with no JavaScript listener, so looking at the event region alone would classify it as NotTracking. HTMLInputElement::updateTouchEventHandler() now sets the HasInternalTouchEventHandling flag on EventTarget for range inputs, and StyleAdjuster turns that flag into the full set of touch region types for the element.

Keeping the animation running on the scrolling thread #

The last piece is in ScrollingEffectsController::handleWheelEvent(). As shown above, WPE synthesizes wheel events from touch gestures with precise deltas. Precise-delta events only need immediateScrollBy() to move the scroll position — but then nothing drives screen updates while the main thread is busy.

The fix is that, while a scroll gesture is in progress, a scroll animation is also started — from one ULP short of the destination (std::nextafter()) to the destination. Visually it finishes instantly — the real scroll is still done by immediateScrollBy() — but a scroll animation is now running, which starts display link monitoring and keeps compositing driven regardless of the main thread.

The event flow, summarized #

Before:

The UI process sends the touch event to the web process main thread, which may
be blocked by JavaScript or layout. Only after the reply arrives does gesture
recognition synthesize a wheel event and scrolling
start.

After (no listeners, or passive listeners only):

The UI process sends the touch event to the EventDispatcher thread of the web
process, which asks the scrolling tree and replies immediately without the main
thread, so gesture recognition synthesizes a wheel event and scrolling starts
right away. If passive listeners exist, the event is also delivered to the main
thread afterwards.

Where a non-passive listener exists, we still wait for the main thread as before. The spec requires preventDefault() to be honoured, so that is unavoidable.

Layout test updates #

As a side effect, the tests under fast/events/touch/ had to be updated.

Event regions are computed during a rendering update and propagated to the scrolling tree via the platform layer tree. Which means a test like this:

target.addEventListener("touchstart", handler);
tapSoon(20, 20); // ← the region has not been updated yet!

taps immediately after registering the listener, while the scrolling tree still believes there is no listener and returns NotTracking. The event never reaches the main thread and the test fails.

A new UIHelper.renderingComplete() was added for this:

static async renderingComplete()
{
// Wait for the platform layer tree to be updated
await UIHelper.animationFrame();
await UIHelper.animationFrame();
}

Two animation frames are needed because the first one runs the rendering update that computes the regions, and a second is needed for the result to reach the layer tree.

Summary #

  • On WPE and GTK both the layer tree and the scrolling tree live in the web process (Coordinated Graphics), so the iOS touch asynchronous scrolling implementation could not be reused directly.
  • Instead, touch events were given the same shape that already works for wheel events: ask the scrolling tree from the EventDispatcher thread.
  • The keys were enabling touch event regions, and making the IPC reply AnyThread so it can be sent without waiting for the main thread.
  • Anywhere the page has no non-passive listener, scrolling now starts regardless of what the main thread is doing.

Acknowledgements #

Many thanks to Alejandro G. Castro and Carlos Garcia Campos for their insightful reviews of this work, and to Claude for writing this blog post.

September 13, 2026 12:00 AM

September 08, 2026

Pawel Lampe: Trying WPE Platform API on Raspberry Pi

Igalia WebKit

WPE Platform API (also known as “new API”) is a redesigned, GObject-based platform-integration layer for WPE WebKit that replaces the older libwpe backend model. A few months ago, Kate and Simon published two closely related blog posts about it. The first one focuses more on the API and browser implementation details, while the second one focuses more on writing and integrating the browser within Linux distribution.

This article builds on top of the above ones, and showcases how to build and try a minimal WPE browser using WPE Platform API on Raspberry Pi. Moreover, as the WPE Platform API still evolves to some degree, this article also explains how to use and stick to the latest WPE WebKit from main branch. This way one can play with all the latest features straight on embedded hardware.

Before going further, one should be aware that in case of a simple release build (instead of one using latest main branch) it’s better to follow official instructions instead of this article.

Setup #

This article focuses on a certain setup using Raspberry Pi 3B but it should be fairly easy to adapt the config to any other Raspberry Pi model.

As for the work environment: the Linux-based host with ability to run containers was used along with WebKit Container SDK. The SDK version was precisely 2.53-v6-d535e88 as it uses Ubuntu 24.04.4 LTS that works well with Yocto scarthgap.

The Yocto scarthgap has been used to increase the chances that the config and commands demonstrated in this article will remain buildable for many years to follow.

Preparing the image #

The preparation of the image starts with a series of commands that create a main directory and clone important Yocto repositories along with some meta layer repositories. At this point already, it’s important to have the working directory shared between host and SDK.

# host
mkdir wpe-upstream
cd wpe-upstream
git clone https://git.yoctoproject.org/git/poky -b scarthgap
git clone git@github.com:openembedded/meta-openembedded.git -b scarthgap
git clone https://git.yoctoproject.org/git/meta-raspberrypi -b scarthgap
git clone https://github.com/Igalia/meta-webkit -b scarthgap
source poky/oe-init-build-env build

Once the build directory is created, it’s necessary to configure the meta layers in the build/conf/bblayers.conf file the following way:

# POKY_BBLAYERS_CONF_VERSION is increased each time build/conf/bblayers.conf
# changes incompatibly
POKY_BBLAYERS_CONF_VERSION = "2"

BBPATH = "${TOPDIR}"
BSPDIR := "${@os.path.abspath(os.path.dirname(d.getVar('FILE', True)) + '/../..')}"

BBFILES ?= ""
BBLAYERS ?= " \
${BSPDIR}/poky/meta \
${BSPDIR}/poky/meta-poky \
${BSPDIR}/poky/meta-yocto-bsp \
${BSPDIR}/meta-openembedded/meta-oe \
${BSPDIR}/meta-openembedded/meta-multimedia \
${BSPDIR}/meta-openembedded/meta-python \
${BSPDIR}/meta-raspberrypi \
${BSPDIR}/meta-webkit \
"

With the above, the recipes from the meta layers cloned earlier will be considered by bitbake.

Next, the most important configuration step is appending the following to build/conf/local.conf:

MACHINE = "raspberrypi3-64" 
MACHINE_FEATURES:append = " vc4graphics"
GPU_MEM_256 = "128"
GPU_MEM_512 = "196"
GPU_MEM_1024 = "396"
DISTRO_FEATURES:append = " opengl egl wayland"
EXTRA_IMAGE_FEATURES = "debug-tweaks"
IMAGE_FEATURES:append = " ssh-server-dropbear hwcodecs"
IMAGE_INSTALL:append = " wpewebkit wpe-browser"
PREFERRED_VERSION_wpewebkit = "latest"
LICENSE_FLAGS_ACCEPTED = "synaptics-killswitch"

With that, wpewebkit latest will be preferred and installed in the image along with a dummy browser called wpe-browser.

To make the wpewebkit latest work, one needs to create meta-webkit/recipes-browser/wpewebkit/wpewebkit_latest.bb:

SUMMARY = "Lightweight WebKit port for embedded devices with OpenGL-ES acceleration"
DESCRIPTION = "WPE WebKit port pairs the WebKit engine with OpenGL-ES (OpenGL for Embedded Systems), \
allowing embedders to create simple and performant systems based on Web platform technologies. \
It is designed with hardware acceleration in mind, relying on EGL, and OpenGL ES."

HOMEPAGE = "https://wpewebkit.org/"
BUGTRACKER = "https://bugs.webkit.org/"
LICENSE = "BSD-2-Clause & LGPL-2.0-or-later"
LIC_FILES_CHKSUM = "file://Source/WebCore/LICENSE-LGPL-2.1;md5=a778a33ef338abbaf8b8a7c36b6eec80 "

REQUIRED_DISTRO_FEATURES = "opengl"

DEPENDS:append = " \
libsoup \
bison-native gperf-native harfbuzz-native libxml2-native ccache-native ninja-native ruby-native \
fontconfig freetype glib-2.0 harfbuzz icu jpeg pcre sqlite3 zlib libpng libtasn1 \
libwebp libxml2 libxslt virtual/egl virtual/libgles2 libepoxy libgcrypt \
unifdef-native \
"


inherit cmake features_check pkgconfig perlnative python3native

export WK_USE_CCACHE = "NO"

PACKAGECONFIG ??= "accessibility avif dfg-jit gbm gpu-process \
jit jpegxl libbacktrace \
mediasource mediastream \
remote-inspector \
sysprof \
${@' system-sysprof' \
if bb.utils.contains('BBFILE_COLLECTIONS', 'meta-gnome', True, False, d) \
else '' }
\
unified-builds video webaudio woff2 wpe-platform \
${@bb.utils.contains('DISTRO_FEATURES', 'systemd', 'journald', '' ,d)} \
"


PACKAGECONFIG[reduce-size] = "-DCMAKE_BUILD_TYPE=MinSizeRel,-DCMAKE_BUILD_TYPE=Release,,"
PACKAGECONFIG[release-with-debug-info] = "-DCMAKE_BUILD_TYPE=RelWithDebInfo,-DCMAKE_BUILD_TYPE=Release,,"

# WPE features
PACKAGECONFIG[accessibility] = "-DUSE_ATK=ON,-DUSE_ATK=OFF,atk at-spi2-atk"
PACKAGECONFIG[avif] = "-DUSE_AVIF=ON,-DUSE_AVIF=OFF,libavif"
PACKAGECONFIG[bubblewrap] = "-DENABLE_BUBBLEWRAP_SANDBOX=ON -DBWRAP_EXECUTABLE=${bindir}/bwrap -DDBUS_PROXY_EXECUTABLE=${bindir}/xdg-dbus-proxy,-DENABLE_BUBBLEWRAP_SANDBOX=OFF,bubblewrap xdg-dbus-proxy libseccomp"
PACKAGECONFIG[developer-mode] = "-DDEVELOPER_MODE=ON,-DDEVELOPER_MODE=OFF,wayland-native wayland-protocols wpebackend-fdo"
PACKAGECONFIG[deviceorientation] = "-DENABLE_DEVICE_ORIENTATION=ON,-DENABLE_DEVICE_ORIENTATION=OFF,"
PACKAGECONFIG[dfg-jit] = "-DENABLE_DFG_JIT=ON,-DENABLE_DFG_JIT=OFF,"
PACKAGECONFIG[documentation] = "-DENABLE_DOCUMENTATION=ON,-DENABLE_DOCUMENTATION=OFF, gi-docgen-native gi-docgen"
PACKAGECONFIG[encryptedmedia] = "-DENABLE_ENCRYPTED_MEDIA=ON,-DENABLE_ENCRYPTED_MEDIA=OFF,libgcrypt"
PACKAGECONFIG[experimental-features] = "-DENABLE_EXPERIMENTAL_FEATURES=ON,-DENABLE_EXPERIMENTAL_FEATURES=OFF,libavif libjxl"
PACKAGECONFIG[gamepad] = "-DENABLE_GAMEPAD=ON,-DENABLE_GAMEPAD=OFF,libmanette"
PACKAGECONFIG[gbm] = "-DUSE_GBM=ON,-DUSE_GBM=OFF,libdrm"
PACKAGECONFIG[geolocation] = "-DENABLE_GEOLOCATION=ON,-DENABLE_GEOLOCATION=OFF,geoclue"
PACKAGECONFIG[gpu-process] = "-DENABLE_GPU_PROCESS=ON,-DENABLE_GPU_PROCESS=OFF,"
PACKAGECONFIG[hyphen] = "-DUSE_LIBHYPHEN=ON,-DUSE_LIBHYPHEN=OFF,hyphen"
PACKAGECONFIG[introspection] = "-DENABLE_INTROSPECTION=ON,-DENABLE_INTROSPECTION=OFF, gobject-introspection-native"
PACKAGECONFIG[jit] = "-DENABLE_JIT=ON -DENABLE_C_LOOP=OFF,-DENABLE_JIT=OFF -DENABLE_C_LOOP=ON,"
PACKAGECONFIG[jpegxl] = "-DUSE_JPEGXL=ON,-DUSE_JPEGXL=OFF,libjxl"
PACKAGECONFIG[journald] = "-DENABLE_JOURNALD_LOG=ON,-DENABLE_JOURNALD_LOG=OFF,"
PACKAGECONFIG[lcms] = "-DUSE_LCMS=ON,-DUSE_LCMS=OFF,"
PACKAGECONFIG[spellcheck] = "-DENABLE_SPELLCHECK=ON,-DENABLE_SPELLCHECK=OFF,enchant"
PACKAGECONFIG[wpe-legacy-api] = "-DENABLE_WPE_LEGACY_API=ON,-DENABLE_WPE_LEGACY_API=OFF,libwpe virtual/wpebackend,"
PACKAGECONFIG[libbacktrace] = "-DUSE_LIBBACKTRACE=ON,-DUSE_LIBBACKTRACE=OFF,libbacktrace"
PACKAGECONFIG[minibrowser] = "-DENABLE_MINIBROWSER=ON,-DENABLE_MINIBROWSER=OFF,wayland-native wayland-protocols wpebackend-fdo"
PACKAGECONFIG[mediasource] = "-DENABLE_MEDIA_SOURCE=ON,-DENABLE_MEDIA_SOURCE=OFF,gstreamer1.0 gstreamer1.0-plugins-good"
PACKAGECONFIG[mediastream] = "-DENABLE_MEDIA_STREAM=ON,-DENABLE_MEDIA_STREAM=OFF,gstreamer1.0 gstreamer1.0-plugins-bad"
PACKAGECONFIG[pdfjs] = "-DENABLE_PDFJS=ON,-DENABLE_PDFJS=OFF,"

PACKAGECONFIG[speech-synthesis] = "-DENABLE_SPEECH_SYNTHESIS=ON,-DENABLE_SPEECH_SYNTHESIS=OFF,flite"

PACKAGECONFIG[sysprof] = "-DUSE_SYSPROF_CAPTURE=ON, -DUSE_SYSPROF_CAPTURE=OFF,"
PACKAGECONFIG[system-sysprof] = "-DUSE_SYSTEM_SYSPROF_CAPTURE=ON, -DUSE_SYSTEM_SYSPROF_CAPTURE=OFF, sysprof"
PACKAGECONFIG[video] = "-DENABLE_VIDEO=ON,-DENABLE_VIDEO=OFF,gstreamer1.0 gstreamer1.0-plugins-base"
PACKAGECONFIG[webaudio] = "-DENABLE_WEB_AUDIO=ON,-DENABLE_WEB_AUDIO=OFF,gstreamer1.0 gstreamer1.0-plugins-base gstreamer1.0-plugins-good"
PACKAGECONFIG[woff2] = "-DUSE_WOFF2=ON,-DUSE_WOFF2=OFF,woff2"
PACKAGECONFIG[remote-inspector] = "-DENABLE_REMOTE_INSPECTOR=ON,-DENABLE_REMOTE_INSPECTOR=OFF,"
PACKAGECONFIG[webrtc] = "-DENABLE_WEB_RTC=ON,-DENABLE_WEB_RTC=OFF,libvpx libevent libopus openh264"
PACKAGECONFIG[qtwpe] = "-DENABLE_WPE_QT_API=ON ${CMAKE_QT_OECONF},-DENABLE_WPE_QT_API=OFF,qtbase-native qtbase qtdeclarative libepoxy wpebackend-fdo ${QT_BUILD_DEPS}"
PACKAGECONFIG[unified-builds] = "-DENABLE_UNIFIED_BUILDS=ON,-DENABLE_UNIFIED_BUILDS=OFF,"
PACKAGECONFIG[thunder] = "-DENABLE_THUNDER=ON,-DENABLE_THUNDER=OFF,virtual/open-cdm"
PACKAGECONFIG[webxr] = "-DENABLE_WEBXR=ON,-DENABLE_WEBXR=OFF,openxr"

# Build option for WPE API 1.1
PACKAGECONFIG[wpe-1-1-api] = "-DENABLE_WPE_1_1_API:BOOL=ON,-DENABLE_WPE_1_1_API:BOOL=OFF,"

# Build option for WPE platform API
PACKAGECONFIG[wpe-platform] = "-DENABLE_WPE_PLATFORM=ON,-DENABLE_WPE_PLATFORM=OFF,libinput libxkbcommon wayland-native"

EXTRA_OECMAKE = " -DPORT=WPE -G Ninja"

# TODO: documentation and introspection are disabled by default because the are
# causing cross-compiling build errors
# PACKAGECONFIG:append = " ${@bb.utils.contains('DISTRO_FEATURES', 'api-documentation', 'documentation', '' ,d)} introspection"

# If SSE code compiles, assume it runs successfully (it can't actually run
# because of cross compiling)
EXTRA_OECMAKE:append:x86 = " -DHAVE_SSE2_EXTENSIONS_EXITCODE=0"
# Javascript JIT is not supported on ppc/arm/RISCV32/mips64
PACKAGECONFIG:remove:powerpc = "jit"
PACKAGECONFIG:remove:powerpc64 = "jit"
PACKAGECONFIG:remove:powerpc64le = "jit"
PACKAGECONFIG:remove:armv4 = "jit"
PACKAGECONFIG:remove:armv5 = "jit"
PACKAGECONFIG:remove:armv6 = "jit"
PACKAGECONFIG:remove:armv7a = "jit"
PACKAGECONFIG:remove:armv7ve = "jit"
PACKAGECONFIG:remove:riscv32 = "jit"
PACKAGECONFIG:remove:riscv64 = "jit"
PACKAGECONFIG:remove:mipsarchn64 = "jit"
PACKAGECONFIG:remove:mipsarchn32 = "jit"
PACKAGECONFIG:remove:loongarch64 = "jit"

# Javascript JIT is not supported on x86
PACKAGECONFIG:remove:x86 = "jit"

LDFLAGS:append:riscv64 = " -pthread"

FULL_OPTIMIZATION:remove = "-g"

LEAD_SONAME = "libWPEWebKit.so"
PACKAGES =+ "${PN}-web-inspector-plugin ${PN}-qtwpe-qml-plugin"
FILES:${PN} += "${libdir}/wpe-webkit*/injected-bundle/libWPEInjectedBundle.so"
FILES:${PN}-web-inspector-plugin += "${datadir}/wpe-webkit-*/inspector.gresource"
# nooelint: oelint.vars.insaneskip - ignored for convenience. We need to recheck if problem persist
INSANE_SKIP:${PN}-web-inspector-plugin = "dev-so"

# nooelint: oelint.vars.insaneskip - ignored for convenience. We need to recheck if problem persist
INSANE_SKIP:${PN}-qtwpe-qml-plugin = "dev-so"

# JSC JIT on ARMv7 is better supported with Thumb2 instruction set.
ARM_INSTRUCTION_SET:armv7a = "thumb"
ARM_INSTRUCTION_SET:armv7r = "thumb"
ARM_INSTRUCTION_SET:armv7m = "thumb"
ARM_INSTRUCTION_SET:armv7ve = "thumb"

# Extra runtime depends
# nooelint: oelint.vars.dependsordered - ignored for convenience
RDEPENDS:${PN} += "\
${@bb.utils.contains('PACKAGECONFIG', 'remote-inspector', '${PN}-web-inspector-plugin', '', d)} \
${@bb.utils.contains('PACKAGECONFIG', 'gst_gl', 'gstreamer1.0-plugins-base-opengl', '', d)} \
${@bb.utils.contains('PACKAGECONFIG', 'mediasource', 'gstreamer1.0-plugins-good-isomp4', '', d)} \
${@bb.utils.contains('PACKAGECONFIG', 'webaudio', 'gstreamer1.0-plugins-good-wavparse', '', d)} \
${@bb.utils.contains('PACKAGECONFIG', 'video', 'gstreamer1.0-plugins-base-app \
gstreamer1.0-plugins-base-audioconvert \
gstreamer1.0-plugins-base-audioresample \
gstreamer1.0-plugins-base-gio \
gstreamer1.0-plugins-base-playback \
gstreamer1.0-plugins-base-typefindfunctions \
gstreamer1.0-plugins-base-videoconvertscale \
gstreamer1.0-plugins-base-volume \
gstreamer1.0-plugins-good-audiofx \
gstreamer1.0-plugins-good-audioparsers \
gstreamer1.0-plugins-good-autodetect \
gstreamer1.0-plugins-good-avi \
gstreamer1.0-plugins-good-deinterlace \
gstreamer1.0-plugins-good-interleave \
', '', d)}
\
libgles2 \
"


RDEPENDS:${PN}-web-inspector-plugin += "\
shared-mime-info \
"


# Extra runtime recommends
RRECOMMENDS:${PN} += "\
ca-certificates \
ttf-dejavu-sans \
ttf-dejavu-sans-mono \
ttf-dejavu-serif \
${PN}-qtwpe-qml-plugin \
${@bb.utils.contains('PACKAGECONFIG', 'video', 'gstreamer1.0-plugins-base-meta gstreamer1.0-plugins-good-meta gstreamer1.0-plugins-bad-meta', '', d)} \
"


DEFAULT_PREFERENCE = "-1"

FILESEXTRAPATHS:prepend := "${THISDIR}/${PN}:"

# https://commits.webkit.org/319279@main
PR = "r319279"
SRCREV = "93472ec12ee947b30c7ec3c176ca0e2fb6ba6ace"
SRC_URI = "git://github.com/WebKit/WebKit.git;protocol=https;branch=main \
file://0001-libpas-Only-include-stdatomic.h-when-compiling-with.patch \
file://0002-WebDriver-Guard-LOG_CHANNEL-check-with-LOG_DISABLED.patch \
"

S = "${WORKDIR}/git"

This file is self-contained on purpose; the idea is not to rely on any includes so that unpredictable behavior doesn’t happen in the future.

While the above file contains a lot of interesting details, the most interesting practical part is the last few lines. The SRCREV is set to point to the latest commit from the WebKit main branch (at the time of writing). Along with that comes SRC_URI that specifies two patches required to make WPE WebKit compile.

The problem with the patches is that advancing SRCREV will likely make them unusable due to merge conflicts. Therefore, when advancing the revision, it’s recommended to remove patches from SRC_URI and face the compilation problems from scratch as it’s very likely there will be new compilation problems anyway. Fortunately, nowadays LLMs can be used to fix any compilation problems by preparing custom patches just like the below ones (created by LLM as well). For example, most of the modern models should manage to prepare proper patches just by pointing them to the above recipe and the temp directory with the latest logs from build commands.

Assuming one uses the wpewebkit_latest.bb above, the first patch needs to be created in: meta-webkit/recipes-browser/wpewebkit/wpewebkit/0001-libpas-Only-include-stdatomic.h-when-compiling-with.patch with the following content:

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Pawel Lampe <plampe@igalia.com>
Date: Mon, 17 Aug 2026 00:00:00 +0000
Subject: [PATCH] [libpas] Only include stdatomic.h when compiling with Clang

Some libpas .c files (e.g. jit_heap.c) are compiled as C++ via
set_source_files_properties(... PROPERTIES LANGUAGE CXX) in
Source/bmalloc/CMakeLists.txt, for TZone heap support. pas_utils.h
unconditionally does `#include <stdatomic.h>`, but GCC's own
<stdatomic.h> relies on the C11-only `_Atomic` keyword and has no
support for being included from C++ translation units (this was only
addressed in much newer GCC releases). Compiling any of the
CXX-tagged libpas .c files with GCC 13 therefore fails with:

    error: '_Atomic' does not name a type

The header is only actually needed here for the Clang-specific
__c11_atomic_* intrinsics guarded by `#elif PAS_COMPILER(CLANG)`
further down in this file; the non-Clang (GCC) path uses the
__atomic_* builtins instead and does not need any of the types or
macros from <stdatomic.h>. Guard the include accordingly so GCC
builds (both plain C and the CXX-tagged libpas sources) are
unaffected by GCC's non-C++-aware <stdatomic.h>.

Upstream-Status: Pending
Signed-off-by: Pawel Lampe <plampe@igalia.com>
---
 Source/bmalloc/libpas/src/libpas/pas_utils.h | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/Source/bmalloc/libpas/src/libpas/pas_utils.h b/Source/bmalloc/libpas/src/libpas/pas_utils.h
index 962634080930..9b98fda54f43 100644
--- a/Source/bmalloc/libpas/src/libpas/pas_utils.h
+++ b/Source/bmalloc/libpas/src/libpas/pas_utils.h
@@ -42,7 +42,15 @@
 #endif
 
 #include <limits.h>
+#if PAS_COMPILER(CLANG)
+/* GCC's <stdatomic.h> relies on the C-only _Atomic keyword and is not
+ * usable when this header is included from a translation unit compiled
+ * as C++ (some libpas .c files are compiled as C++, see bmalloc's
+ * CMakeLists.txt). It is only needed here for the Clang-specific
+ * __c11_atomic_* intrinsics below; the GCC path uses __atomic_* builtins
+ * instead. */
 #include <stdatomic.h>
+#endif
 #include <stdbool.h>
 #include <stdint.h>
 #include <string.h>
--
2.43.0


The second patch should be: meta-webkit/recipes-browser/wpewebkit/wpewebkit/0002-WebDriver-Guard-LOG_CHANNEL-check-with-LOG_DISABLED.patch with:

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Pawel Lampe <plampe@igalia.com>
Date: Mon, 17 Aug 2026 00:00:00 +0000
Subject: [PATCH] [WebDriver] Guard LOG_CHANNEL check with
 LOG_DISABLED/RELEASE_LOG_DISABLED

WebDriverService::handleRequest() unconditionally checks
LOG_CHANNEL(WebDriverClassic).state to decide whether it is worth
building the request/response log strings. However,
Source/WebDriver/Logging.h only declares the WebDriverClassic (and
other WebDriver) log channels inside:

    #if !LOG_DISABLED || !RELEASE_LOG_DISABLED

On a release build (NDEBUG, so LOG_DISABLED is true) without journald
support and without OS_LOG/Android (so RELEASE_LOG_DISABLED is also
true) - the common configuration for an embedded Linux build without
the "journald" PACKAGECONFIG - that guard is false, so the channel is
never declared, and this direct, unguarded use of LOG_CHANNEL() fails
to compile:

    error: 'LOG_CHANNEL_PREFIXWebDriverClassic' was not declared in this scope

RELEASE_LOG_INFO() itself already collapses to a no-op in that
configuration (see wtf/Assertions.h), so guard this manual
LOG_CHANNEL() state check with the same condition used to declare the
channel in Logging.h.

Upstream-Status: Pending
Signed-off-by: Pawel Lampe <plampe@igalia.com>
---
 Source/WebDriver/WebDriverService.cpp | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/Source/WebDriver/WebDriverService.cpp b/Source/WebDriver/WebDriverService.cpp
index 07ae385f33f4..2612d239a611 100644
--- a/Source/WebDriver/WebDriverService.cpp
+++ b/Source/WebDriver/WebDriverService.cpp
@@ -381,6 +381,7 @@ bool WebDriverService::findCommand(HTTPMethod method, const String& path, Comman
 void WebDriverService::handleRequest(HTTPRequestHandler::Request&& request, Function<void (HTTPRequestHandler::Response&&)>&& replyHandler)
 {
     Function<void (HTTPRequestHandler::Response&&)> actualReplyHandler = WTF::move(replyHandler);
+#if !LOG_DISABLED || !RELEASE_LOG_DISABLED
     if (LOG_CHANNEL(WebDriverClassic).state != WTFLogChannelState::Off) {
         RELEASE_LOG_INFO(WebDriverClassic, "HTTP request %s %s (body=%zu bytes)", request.method.utf8().data(), request.path.utf8().data(), request.dataLength);
         actualReplyHandler = [startTime = MonotonicTime::now(), replyHandler = WTF::move(actualReplyHandler)](HTTPRequestHandler::Response&& response) mutable {
@@ -388,6 +389,7 @@ void WebDriverService::handleRequest(HTTPRequestHandler::Request&& request, Func
             replyHandler(WTF::move(response));
         };
     }
+#endif
 
     auto method = toCommandHTTPMethod(request.method);
     if (!method) {
--
2.43.0

Once the patches are added, wpewebkit latest should build correctly. However, to make use of it one needs a browser.

To demonstrate how easy the browser for WPE WebKit with Platform API can be, a very minimalistic one will be prepared below.

The first step is to create a directory:

mkdir meta-webkit/recipes-browser/wpe-browser/

Then a file: meta-webkit/recipes-browser/wpe-browser/main.cpp that implements the whole browser:

#include <wpe/webkit.h>

int main(int argc, const char *argv[]) {
g_autoptr(GMainLoop) loop = g_main_loop_new(nullptr, false);
g_autoptr(WebKitWebView) view = WEBKIT_WEB_VIEW(g_object_new(WEBKIT_TYPE_WEB_VIEW,
nullptr));
webkit_web_view_load_uri(view,
(argc > 1) ? argv[1] : "https://wpewebkit.org");
g_main_loop_run(loop);
return EXIT_SUCCESS;
}

Then a file: meta-webkit/recipes-browser/wpe-browser/CMakeLists.txt that describes how to build it:

cmake_minimum_required(VERSION 3.16)
project(wpe-browser CXX)

set(CMAKE_CXX_STANDARD 17)

include(GNUInstallDirs)

find_package(PkgConfig REQUIRED)

# The Wayland WPE Platform already depends on wpe-platform-2.0
pkg_check_modules(WebKitDeps REQUIRED
IMPORTED_TARGET
wpe-webkit-2.0
wpe-platform-wayland-2.0
)

add_executable(wpe-browser main.cpp)

target_link_libraries(wpe-browser
PRIVATE
PkgConfig::WebKitDeps
)

install(TARGETS wpe-browser RUNTIME DESTINATION ${CMAKE_INSTALL_BINDIR})

and finally a recipe file: meta-webkit/recipes-browser/wpe-browser/wpe-browser_1.0.bb that allows bitbake to build the browser and install it in the image:

SUMMARY = "Minimal WPE WebKit browser launcher"
DESCRIPTION = "A minimal launcher built on the WPE Platform API, displaying \
a URL given as its only argument (defaults to https://wpewebkit.org). \
Based on https://simonpena.com/blog/2026/03/20/getting-started-with-wpe-webkit/"

HOMEPAGE = "https://simonpena.com/blog/2026/03/20/getting-started-with-wpe-webkit/"
LICENSE = "MIT"
LIC_FILES_CHKSUM = "file://${COMMON_LICENSE_DIR}/MIT;md5=0835ade698e0bcf8506ecda2f7b4f302"

FILESEXTRAPATHS:prepend := "${THISDIR}:"

SRC_URI = "file://main.cpp \
file://CMakeLists.txt \
"


S = "${WORKDIR}"

DEPENDS = "wpewebkit"
RDEPENDS:${PN} += "wpewebkit"

inherit cmake pkgconfig features_check

REQUIRED_DISTRO_FEATURES = "opengl wayland"

After all those steps, everything is ready to perform a build. This time the commands are executed in the SDK:

source poky/oe-init-build-env build
bitbake core-image-weston
cd tmp/deploy/images/raspberrypi3-64/
# flash to SD card based on preference

Trying the image #

Once the image is built and flashed to SD card, and the SD card has been used to boot the Raspberry Pi, one can SSH into it, basically by:

ssh root@<IP>

Then, the browser should work out of the box. The first command to try is the one that doesn’t need the network and therefore is the simplest:

XDG_RUNTIME_DIR=/run/user/1000 WAYLAND_DISPLAY=wayland-1 wpe-browser 'webkit://gpu/stdout'

If the network is available, it’s worth starting simple and loading some HTTP page:

XDG_RUNTIME_DIR=/run/user/1000 WAYLAND_DISPLAY=wayland-1 wpe-browser 'http://info.cern.ch'

If that works as well, one can try the HTTPS one:

XDG_RUNTIME_DIR=/run/user/1000 WAYLAND_DISPLAY=wayland-1 wpe-browser 'https://igalia.com'

If there’s an issue with certificates, it’s likely due to broken date/time, so the correct one needs to be set using a command like:

date -s '2026-09-02 22:34:56'

Conclusions #

Since the Platform API is the future of WPE, it’s worth using it already. As the above sections demonstrate, it hasn’t ever been easier to create a WPE-powered browser for embedded hardware. Moreover, nowadays as LLMs are at play, using latest sources from main branch and quickly patching on demand is a possibility worth utilizing. With that, a broad sea of experimenting possibilities becomes wide open. However, it must not be forgotten that while main branch is great for testing new web platform features implemented in WebKit, it’s not necessarily ideal for testing performance. In such case, it’s better to rely on releases and proper browser engine fine tuning for particular hardware one plays with.

September 08, 2026 12:00 AM

September 07, 2026

Igalia WebKit Team: WebKit Igalia Periodical #76

Igalia WebKit

Update on what happened in WebKit in the week from August 31 to September 7.

This was another week focused on ironing out graphics issues in preparation for the upcoming 2.54.x release series. Did we say release? Here we have another set of packaged release candidates as well!

Cross-Port 🐱

The “paint flushing” feature of the Web Inspector that shows the areas of the layers that were painted is now available. On the other hand, setting WEBKIT_SHOW_DAMAGE=1 in the environment will show the areas of the window that were rendered. For example, if the page is scrolled a little, the whole window is rendered, but no layer has to be repainted.

Graphics 🖼️

Fixed visible pixel snapping in the GTK and WPE ports' Skia compositor by choosing linear instead of nearest-neighbor sampling for backing store tiles whenever a layer's transform no longer maps tile pixels onto screen pixels 1:1, matching what the Texture Mapper backend already got from its always-linear OpenGL code path.

Fixed a clipping bug in the Skia-based compositor that caused layers with overflow: hidden set to not clip blur and box-shadow filter outsets.

Fixed a repaint bug where content changing behind an element with a software-rendered filter such as blur() could leave stale pixels on screen, as with the glow effect that YouTube draws around videos in dark/ambient mode. The fix restores proper tracking of which layers act as the repaint container for a pixel-moving filter, so the repaint area is expanded correctly again on WPE, GTK and macOS.

A few problems with Offscreen Canvas have been fixed.

Releases 📦️

Release candidates WebKitGTK 2.53.92 and WPE WebKit 2.53.92 have been published, and they include polishing and a number of fixes that ensure that there will not be noticeable regressions brought in by the new Skia-based compositor and the damage tracking support—two of the main features of the upcoming stable release series. As the stable release dates approaches, we encourage people who try these preview versions to report issues in Bugzilla.

That’s all for this week!

By Igalia WebKit Team at September 07, 2026 11:43 PM

August 31, 2026

Igalia WebKit Team: WebKit Igalia Periodical #75

Igalia WebKit

Update on what happened in WebKit in the week from August 24 to August 31.

After such a packed edition last week, we can relax with an even more packed installment this week! The team has been moving full steam ahead with the Layer-Based SVG Engine, and graphics improvements in general, but we also had a handful of other updates, such as SDK updates, a new WebsitePolicies API, and more.

Cross-Port 🐱

Add new ChildChange types for moveBefore(). This addresses issues with the in-progress moveBefore() implementation where scripts would sometimes execute erroneously.

Added alpha channel support to color input choosers.

Fixed Service Worker static routes on non-Apple ports.

Added a WebsitePolicies:upgrade-to-https-policy property. When enabled this policy will automatically try using HTTPS even for HTTP websites (excluding localhost or IPs). The policy can be configured either to allow automatically falling back to HTTP if that fails, or to consider all HTTPS failures fatal for the best security.

Graphics 🖼️

Fixed SVG text being rasterized for the wrong resolution in zoomed standalone SVG documents in the Layer-Based SVG Engine (LBSE). vector-effect: non-scaling-stroke on <text> no longer comes out too thick, and the resulting metrics match the legacy SVG engine.

Fixed viewport clipping in the Layer-Based SVG Engine (LBSE), where a nested <svg> or a <marker> applied its viewport clip even when its content already fitted inside, and since that clip is not pixel-snapped its edge could fall between two device pixels and cut into whatever was drawn right at the viewport border. Painting now skips a clip that removes nothing, which eliminates a class of subtle pixel differences against the legacy SVG engine.

Fixed opacity animations not damaging descendant layers in the GTK and WPE ports, where a descendant that paints outside its parent's bounds kept its stale pixels on screen. The mask-specific damage handling was generalized into a single group-property path shared by opacity, filters, mask blend modes and replicas, which now damages the layer plus the overlap region of its whole subtree.

Added a cycle-analysis subcommand to webkit-sysprof, which draws every frame cycle of a capture as a bar of cells colored by the mark covering that moment on the main thread, showing whether a slow frame stalled on layout, a long timer or rasterization instead of only reporting that it was slow.

Removed redundant text updates for the text children of elements using display: contents, which previously got one on every style resolution regardless of whether their style actually changed. This removes dozens of useless updates per style recalculation in Web Component applications, where every <slot> uses display: contents.

Fixed tile image caching in the Skia compositor by regenerating the cached SkImage whenever a a tile's contents are updated and by adding a texture release callback that keeps the texture valid for as long as the image references it.

Switched video DMA-BUF buffers to Skia promise images in the Skia compositor, so the texture backing a video frame is only created at the point Skia actually draws it, which works for these buffers because the underlying DMABufBuffer can be kept alive until the promise image is released.

Fixed a hang in the Layer-Based SVG Engine (LBSE) when two SVG <pattern> elements reference each other through href, or one references itself: collecting the inherited pattern attributes now remembers which patterns it has already visited, the same cycle detection that gradients have always had. The walk also resolves every reference in the tree scope of the pattern it started from, so a pattern referenced from inside a shadow tree now inherits the right attributes.

Skipped anchor-positioning bookkeeping during style resolution when a document uses no anchor positioning at all, avoiding two hash lookups per styled element.

Cached the SVG viewport size used to resolve lengths in the Layer-Based SVG Engine (LBSE), instead of recomputing the nearest <svg> element's view box rectangle once per shape per frame. The viewport is invariant across a flush and identical for every shape under the same <svg>, so caching removes redundant work.

Removed the code path for GPU rendering without Deferred Display Lists (DDL) in the Skia compositor, so accelerated painting always records into a display list and no longer needs to create GL contexts on worker threads.

Fixed red and blue appearing swapped in non-accelerated video with the Skia compositor, by allocating the video frame's BitmapTexture with the BGRA layout flag and applying that flag when the buffer is turned into a Skia image.

Infrastructure 🏗️

Bumped the GTK and WPE developer SDK from v9 to v11, bringing GStreamer 1.28.5, sparkle-cdm 2026.2 and libsoup 3.7.2. Be sure to update your local wkdev-sdk container using wkdev-update to make sure your development environment matches what the CI is testing.

That’s all for this week!

By Igalia WebKit Team at August 31, 2026 08:12 PM

August 24, 2026

Igalia WebKit Team: WebKit Igalia Periodical #74

Igalia WebKit

Update on what happened in WebKit in the week from August 17 to August 24.

Another periodical packed with updates on the graphics, multimedia, and tooling fronts. Which sure has to do with preparing for the upcoming 2.54.x release series, that now has release candidates published. Also, do not miss a new stable release being published with fixes for security issues, and make sure to update.

Cross-Port 🐱

Fixed flakiness in the Web Inspector heap snapshot tests.

Extended the webkit-sysprof analyze tool cover the remaining marks emitted along the rendering pipeline, from RenderTreeBuild and CompositingUpdate through FinalizeRenderingUpdate, RenderLayerTree and WaitForCompositionCompletion down to the individual tile marks, pulling the tile count and the dirty region out of the mark messages as statistics next to the durations. The statistics tables also gained mean and median columns, so a captured trace now shows where the time in a frame actually goes.

Fixed test fast/canvas/canvas-composite-text-alpha.html, and improved the state of several other tests which relied on setTimeout().

Implemented moving steps for <option> elements.

Add support for asynchronous scrolling with touch events.

Touch events are now dispatched to the EventDispatcher thread in the Web Process, enabling smooth scrolling even when the main thread is busy. Additionally, this change fixed a bug where precise scrolling delta wheel events, dispatched by touchpads, failed to trigger asynchronous scrolling.

Multimedia 🎥

GStreamer-based multimedia support for WebKit, including (but not limited to) playback, capture, WebAudio, WebCodecs, and WebRTC.

Ensure that multimedia on pages restored from the back-forward cache is correctly processed without errors or reloading.

The robustness level is now queried from the CDM (if supported) when using Encrypted Media Extensions (EME).

Added video rendering support for Qualcomm hardware-accelerated decoders when the Skia compositor is in use, complementing the earlier TextureMapper-only support that left such videos blank with the new compositor.

Video frame processing now relies on the driver's implicit YUV to RGB conversion, steered by the colour space and sample range hints taken from the frame colorimetry, with a hint-free import as fallback.

Graphics 🖼️

Fixed how <feImage> paints its referenced element in the Layer-Based SVG Engine (LBSE), which still used a helper written for the legacy SVG engine, where SVG content never had layers, so it used to paint renderers directly and skipped any child that owns a RenderLayer under LBSE, silently dropping its opacity, mask, filter or 3D transform. The referenced content is now painted through the layer tree, the way <mask>,<clipPath>, <pattern> and <marker> content already is.

Made SVG mask no longer force a RenderLayer in the Layer-Based SVG Engine (LBSE), mirroring the earlier change for clip-path. A mask is now applied during painting by SVGNonLayerClippingAndMaskingScope, which opens one transparency layer capturing the renderer's foreground and composites the mask over it afterwards, and which also absorbed the clips that cannot be expressed as a path. A container with a mask keeps its layer, since the mask covers its whole subtree, while a leaf stays layer-free. This further reduces the layer overhead that has been holding LBSE back against the legacy SVG engine.

Skipped scroll coordination for composited layers that have no scrolling role, where the per-layer update previously walked every branch to detach roles the layer had never registered for, only to hand back the parent node ID unchanged. Cutting that work out shortens every compositing update, which matters for composition-heavy workloads.

Fixed rendering of an outermost <svg> with an empty viewBox in the Layer-Based SVG Engine (LBSE), which per the SVG specification should disable painting when the width or height is zero, matching what the legacy engine already did.

Skipped scroll coordination work for layers that have no scrolling role during compositing updates, where updateScrollCoordinationForLayer previously walked every branch to detach roles the layer had never registered for, only to hand back the unchanged parent node ID.

Releases 📦️

WebKitGTK 2.52.6 and WPE WebKit 2.52.6 have been released, including a number of fixes for security issues covered in the accompanying security advisory WSA-2026-0005 (GTK, WPE). It is recommended for everybody to update to these stable releases.

Stabilization for the upcoming 2.54.x release series for both the GTK and WPE ports is ongoing, with the first stable release, 2.54.0, expected around mid-September 2026. In the meantime release candidates WebKitGTK 2.53.91 and WPE WebKit 2.53.91 have been released.

Those interested in previewing the work done by the team in the last half year, including the new Skia-based compositor that is expected to eventually replace the aging TextureMapper, may want to give them a try and report any issues found in Bugzilla.

That’s all for this week!

By Igalia WebKit Team at August 24, 2026 11:30 PM

August 17, 2026

Igalia WebKit Team: WebKit Igalia Periodical #73

Igalia WebKit

Update on what happened in WebKit in the week from August 10 to August 17.

Following an extra packed periodical, this week we get back to a more regular pace with two nice bugfixes, and a new tool to analyze WebKit performance on Linux!

Cross-Port 🐱

The webkit-sysprof toolkit landed in main thus introducing a set of tools for processing Sysprof .syscap capture files recorded from WebKit (GTK/WPE ports). It extracts marks (timeline events) and counters (time-series metrics) from a capture and lets one dump, summarize, analyze, or plot delta-time histograms for them.

Graphics 🖼️

Fixed filters specified on the outermost <svg> element in the Layer-Based SVG Engine (LBSE), where a filter: url(...) reference on an SVG root was silently dropped because the layer code skipped it, as the legacy engine used to apply it by itself. The filter region is now resolved against the SVG root's border box in its container's coordinate system, since the outermost <svg> is a replaced element in the CSS box tree, not part of the SVG user space its children live in.

Avoided serializing gradient and pattern transforms just to answer a presence check in the Layer-Based SVG Engine (LBSE). Asking hasAttribute() whether gradientTransform or patternTransform was specified forced the transform list to be serialized into the attribute map whenever the base value was changed through the SVG DOM, even though that string is never read back, so the check is now answered directly from the typed accessor.

That’s all for this week!

By Igalia WebKit Team at August 17, 2026 07:08 PM

August 10, 2026

Igalia WebKit Team: WebKit Igalia Periodical #72

Igalia WebKit

Update on what happened in WebKit in the week from July 28 to August 10.

Quite a packed pair of weeks this time! The range of updates is big, but some highlights are the handful of Layer-Based SVG Engine updates, performance improvements, and the new WPE APIs. Finally, the Web Engines Hackfest recordings are now published!

Cross-Port 🐱

Fixed webkit_website_data_get_size() looking up sizes for localStorage, indexedDB, and the DOM Cache.

Enabled support for the File System and Storage APIs.

Added WEBKIT_WEBSITE_DATA_FILE_SYSTEM API to enable fetching/clearing File System data sites use.

Multimedia 🎥

GStreamer-based multimedia support for WebKit, including (but not limited to) playback, capture, WebAudio, WebCodecs, and WebRTC.

Landed a follow-up fix for the h264 edit list support to prevent regressions on YouTube MSE Conformance Tests 2019 when using older versions of GStreamer.

Avoided a spurious seek to zero when seeking to the end of an audio/video when playback starts.

Graphics 🖼️

Recovered a MotionMark compositing regression in the Skia backend, where respecting damage information during compositing cost about 20% on the composition suite and more than 50% on two of its tests. Restricting a draw to the damaged area means splitting it into source-rect-to-destination-rect pieces or drawing it under a device-space clip, and neither is needed when the damage already covers the whole draw, which is the common case in those tests because a composited layer is much smaller than a damage grid cell. Those draws are now issued exactly as they would be with damage turned off, avoiding a clip path that flushed the image set batch and left it an order of magnitude smaller, and with the regressions gone, using damage information for compositing was enabled again along with unifying damaged regions that are sent to the system compositor.

Skipped a per-frame visual overflow recomputation in the container paint cull of the Layer-Based SVG Engine (LBSE). The cull used a cached overflow rect that was recomputed by unioning all descendant bounds on a miss, which happened every frame for containers with an animated transform. It now only runs when the rect is already cached.

Skipped the outline paint pass for SVG renderers without an outline in the Layer-Based SVG Engine (LBSE). Every shape used to be painted twice per frame, the second pass being a no-op in the common outline-free case, so guarding it behind hasOutline() removes a redundant traversal from every frame.

Fixed masked SVG content being cut off at the edges in the Layer-Based SVG Engine (LBSE), where the mask image was sized over the enclosing integer rect of the mask content bounds in device space while the transparency layer clip was computed differently, losing the outermost pixels. An SVG renderer that is a box (<text>, <foreignObject>) also used the CSS mask clip rect derived from the border box, which leaves out SVG content spilling outside it, and now uses the visual overflow rect instead

Fixed most of the remaining <mask> issues in the Layer-Based SVG Engine (LBSE): masks were displaced on targets whose children carry transforms, the mask region given by x, y, width and height was ignored so content reaching past it was not cut off, and the cached mask image was never dropped on layout, leaving a resized viewport masking with an image rasterized for the old size.

Stopped rebuilding objectBoundingBox gradients on every layout size change in the Layer-Based SVG Engine (LBSE), which used to discard the cached gradient and re-collect its attributes (serializing every animated property back to a string, including gradientTransform) on the next paint. That work is wasted for objectBoundingBox units, whose coordinates resolve against the object bounding box with the userspace transform recomputed on every paint anyway, so only the clients are repainted now, while userSpaceOnUse gradients resolve against the viewport and are still invalidated as before.

Cached the SVG fill and stroke paint server directly on the renderer in the Layer-Based SVG Engine (LBSE), instead of in the shared referenced-resources table living in a renderer's rar data, where every fill and every stroke paid the cost of a hash map lookup just to reach the cache, which gives a small win on the MotionMark/Suits performance test. It also fixed a shape referencing a paint server that does not exist yet, which kept painting unfilled once an element finally took that id, because the shape registered itself as a pending resource under the full resolved URL while the lookups used the bare fragment identifier.

Sped up mapping the paint dirty rect through transforms in the Layer-Based SVG Engine (LBSE). Transformed SVG paints inverted the full 4x4 matrix on every paint, and now use the cheaper inverse of the 2x3 affine transform whenever the transform is affine, falling back to the 4x4 inverse only for 3D transforms.

Made SVG clip-path no longer force a RenderLayer in the Layer-Based SVG Engine (LBSE). A bare clip is now applied during painting through a shared ClipPathPaintScope, a scope object that sets up the clip in its constructor and tears it down afterwards, handling CSS basic-shape and box clips as well as SVG clipper resources so both regular CSS boxes and SVG content share one path. This is a further step in removing the intrinsic need for layers on SVG renderers, continuing the effort to close the performance gap between LBSE and the legacy SVG engine.

Fixed dynamic x and y updates on SVG <foreignObject> elements, which stopped taking effect after the viewport geometry started being derived from the resolved style. The x, y, width and height attributes are presentation attributes mapped to the CSS x, y, width and height properties, but only width and height marked the presentational hint style as dirty when they changed, so a style recalc never ran for x and y and layout kept reading stale values. All four geometry attributes now invalidate the presentational hint style, matching how <rect> handles its geometry, so setting x.baseVal.value from script repositions the <foreignObject> as expected.

Added support for external and data: URL references to clip-path, markers and paint servers (gradients and patterns) in the Layer-Based SVG Engine (LBSE). Until now the LBSE resource resolvers only looked for the referenced fragment inside the local document, so markup like url(file.svg#id) silently resolved to nothing, while filters already worked and the legacy SVG engine handled all of these since a few weeks.

Fixed SVG filters vanishing on elements with a very large bounding box in the Layer-Based SVG Engine (LBSE). The filter region was seeded with the element's object bounding box and then united with each referenced <filter> region, but a referenced <filter> brings its own region, and that region alone decides where the filter paints, so the union could grow far past the image buffer limits and get clamped down to scale that made the output disappear. When every function in the chain is a <filter> reference the bounding box is now dropped and only the referenced regions are kept, matching what the legacy SVG engine does, while objectBoundingBox filter units still resolve exactly as before and HTML/CSS filters are untouched.

WPE WebKit 📟

Add the WebView::run-color-chooser API to WPE to allow applications to show color choosers, similar to WebKitGTK's API.

The WPE port can now use libsecret for persistent credential storage, reusing the implementation from the WebKitGTK port. This is disabled by default and can be toggled passing -DUSE_LIBSECRET=ON to CMake when configuring the build.

Add the WebKitClipboardPermissionRequest API to WPE, allowing support for the clipboard permission similar to WebKItGTK.

Community & Events 🤝

The videos of the Web Engines Hackfest 2026 talks have been published, including the sessions from the new WPE WebKit track. This year the following WebKit-related talks have been recorded:

That’s all for this week!

By Igalia WebKit Team at August 10, 2026 06:51 PM

July 28, 2026

Igalia WebKit Team: WebKit Igalia Periodical #71

Igalia WebKit

Update on what happened in WebKit in the week from July 14 to July 27.

This two-week update includes plenty of changes to the Skia compositor, changes to multimedia support, three blog posts, and assorted improvements.

Cross-Port 🐱

The Web Inspector “Layout & Rendering” timeline now shows a Layout Invalidated event for every element that needs relayout, not just the layout root (with the old root-only event renamed to Layout Scheduled). This unveils why some layouts take much longer than others. No more guessing which of dozens of nodes is actually to blame!

The webkit://gpu page has gained a dark style, which will be used when the system settings indicate that dark mode is preferred by the user.

Multimedia 🎥

GStreamer-based multimedia support for WebKit, including (but not limited to) playback, capture, WebAudio, WebCodecs, and WebRTC.

The experimental GstWebRTC backend was removed and libwebrtc usage was enabled in the main branch. We hope to enable WebRTC support by default in the 2.56 series, scheduled around March 2027.

MP4 edit lists support was enabled in the MSE backend, improving timestamp accuracy, specially when handling of B-frames.

Graphics 🖼️

Split the compositing walk in the Skia compositor into a damage pass and a paint pass, so the frame damage is known before the first draw. The damage pass walks the layer tree with a SkNoDrawCanvas in place of the real canvas, so every draw is discarded and only the damage is collected. Both passes run from a single paint() that applies animations and computes the transforms once, so the two see the same tree. Knowing the damage up front is what lets the compositor eventually paint only the parts of a frame that actually changed.

Wired up damage-driven compositing on the Skia compositor, so a frame re-composites only the region that actually changed instead of the whole surface, when the UseDamagingInformationForCompositing feature is enabled (not yet on by default). Each frame's damage is combined with what each swap-chain target still needs to redraw since it was last drawn into, and the clear and every draw are clipped to that region, which is a milestone towards no longer repainting untouched pixels every frame.

Made the root layer collect the frame damage itself in the Skia compositor, instead of having each layer report its own changes. Reporting leaves a gap whenever a layer is in no position to report, e.g. a destroyed one took its painted rectangle with it, so what it had drawn stayed on screen. The root now holds one rectangle per layer and compares it against what each frame's walk finds, so a layer that moved is repainted in both places, and a layer the walk never reaches is repainted where it used to be and dropped. Nothing has to notice anything for the pixels it left behind to be repainted, which is what makes it safe to restrict composition to the damaged region by default in future commits.

Limited every content draw to the target's repaint region in the Skia compositor, so a composited frame can redraw only the pixels that actually changed. Each content type restricts itself to the region's rectangles rather than clipping the canvas, since a multi-rectangle clip cannot be a hardware scissor and would make Skia build a mask and break batching. This is the groundwork for damage-driven compositing, which stays off by default behind the damage-tracking feature flag, as the compositor still passes no region and nothing is restricted yet.

Made each swap-chain target track its own damage since it was last current. Repainting only what changed is correct only when drawing into the target that holds the previous frame, but the swap chain hands back whichever target is free, and that one is a frame or more behind. Each frame's damage is now added to every target as it is recorded and cleared from a target when that target is presented, instead of being built as a side effect of reading it.

Taught the tile and image draws in the Skia compositor to split themselves by damage rectangle, so a frame only repaints the parts of a layer that actually changed. A new SkiaDamageRegion holds the frame's damage in device space and is built once per frame, and each draw is restricted to it: skipped when it touches no damage, split into one sub-draw per damage rectangle it overlaps, or drawn under a device-space clip when a rotated or skewed transform rules out working with rectangles. Nothing feeds a damage region in yet, so every draw still paints in full—this prepares for future patches enabling using damage information in the composition

Fixed missing repaints when compositor-applied layer state changes dynamically in the Coordinated Graphics backend. A layer recorded damage when its backing store re-rendered or a new contents buffer arrived, but the compositor also handles filters, masks, clip path changes, the contents rectangle, the contents tiling, the blend mode and contents visibility, and changing any of those alters the pixels it produces without dirtying a tile. Those setters now damage the whole layer, so a compositor that repaints only the damaged rectangles no longer leaves the previous frame's pixels on screen.

Community & Events 🤝

Nikolas Zimmermann has written a two-part blog series about the current the new Layer-Based SVG Engine (LBSE), with the first post covering the effort to reduce layer overhead using layers conditionally, and the second about how compositing is being implemented and the complications introduced due to paint ordering rules.

Loïc Le Page has published a blog post explaining how to use the new WPEPlatform API to implement a custom WPE integration. While presented example uses GLFW and EGL to show Web content on an X11 window, the concepts are useful for anyone looking into embedding WPE.

That’s all for this week!

By Igalia WebKit Team at July 28, 2026 01:04 AM