`InlineFormattingContextLayout::finish_current_line_and_reset()` has an
early return in case the line has no fragment. However, if the line only
has a forced line break, then we still need to set the baseline.
Testing: Adding new test.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
I think this is the correct solution? GlobalScope::get_cx() doesn't take
a self parameter anymore.
Testing: WebGL tests should suffice.
Fixes: #39228
Signed-off-by: lumiscosity <averyrudelphe@gmail.com>
Make several changes which should address panics and inconsistent
behavior around attempts to set extreme sizes:
1. Limit the minimum size of the `RenderingContext` to 1 pixel by 1
pixel. This should address problems where users of the API try to
directly set the size to a zero or negative dimension. In addition,
improve the documentation around `WebView::resize` to mention this.
2. Clamp values sent in the `WebViewDelegate::request_resize_to` method
to be at least 1x1. This prevents Servo from sending nonsense values
to embedders. Improve documentation in this method.
3. In servoshell:
- More consistently clamp inner and outer window size values.
- Clamp all resize values to the available screen size, so that
large screen sizes aren't processed directly.
Testing: This change fixes an existing WPT and adds two new API tests.
Fixes: #36763.
Fixes: #36841.
Fixes: #39141.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
FxHash is faster than FnvHash and SipHash for simple types up to at
least 64 bytes. The cryptographic guarantees are not needed for any
types changed here because they are simple ids.
This changes the types in script and net crates.
In a future PR we will change the remaining Fnv to be also Fx unless
there is a reason to keep them as Fnv.
Testing: Should not change functionality but unit test and wpt will find
it.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
I hit this many times while working on #38717
Testing: Not needed because we just update the docs
---------
Signed-off-by: Sam <16504129+sagudev@users.noreply.github.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
There are many important changes here:
- Generalize the presentation buffer into standalone staging buffers
that hold their own state. This allow them to be used by getImage.
- Move all clear handling to the ScriptThread and send the configuration
on each request present/getimage, thus avoiding any recreate/clearing
messages. This means that we prepare staging buffers lazily, on the
first request.
Try run for this change:
https://github.com/sagudev/servo/actions/runs/17341982368
Testing: This is covered by existing WebGPU CTS tests. There are some
bad expectations updates, but they are also on main (presumably from
last update the rendering work) although I think CTS is actually wrong
(see https://github.com/gpuweb/cts/issues/4440).
Fixes: #36820Fixes: #37705Fixes: #33368 (we now keep reference alive in hashmap)
---------
Signed-off-by: sagudev <16504129+sagudev@users.noreply.github.com>
Reincorporate the styles to help with the vertical alignment of textual
`<input>` element that was removed by #38775. Followed by adding Servo
specific WPTs to help with these kinds of problem.
Testing: New WPT
Signed-off-by: Jo Steven Novaryo <jo.steven.novaryo@huawei.com>
In servo, each `LocalFontIdentifier` has an `fn index() -> u32`, which
returns the index of the font within the font file in case it is a font
collection (`.ttc` instead of `.ttf`). The way this index is obtained is
platform-dependent.
On systems using `fontconfig`, we get the index by querying `FC_INDEX`:
d2c78db981/components/fonts/platform/freetype/font_list.rs (L109-L112)
There is a sneaky bug here: In addition to the aforementioned face
index, the value for `FC_INDEX` contains the index of the named instance
of the face in the upper 16 bits. This behaviour is completely
undocumented, but FreeType uses the same format:
https://freetype.org/freetype2/docs/reference/ft2-face_creation.html#ft_open_face.
This wasn't a problem for the longest time, because the only consumer of
the `index` value coming from fontconfig was FreeType. However, sometime
after
29e618dcf7,
we started also passing it to`read-fonts` . `read-fonts` expects only
the face index, causing it to return errors as seen in
https://github.com/servo/servo/issues/39209.
The fix is to seperate the two indices when storing them in a
`LocalFontDescriptor` and pass only the lower 16 bits to `read-fonts`.
I'm unsure whether we should continue passing the named instance index
to FreeType since we don't actually support variable fonts, but I've
opted to keep the current behaviour for now.
Fixes: https://github.com/servo/servo/issues/39209
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
This moves more of HashMap/FnvHashMap to FxHashmap. Again we only
changed instances that do not look security related and have small keys.
Additionally, allocator used the fnv feature which did not seem to be
used.
Testing: Unit Tests and WPT should cover this and functionality change
is highly unlikely.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Some animated images in the wild have delays <10ms and browsers usually
change them into 100ms as such small timings are unusual.
Relevant code in FF:
https://searchfox.org/firefox-main/source/image/FrameTimeout.h#35
Testing: Manually tested
Fixes: #39187
Signed-off-by: sagudev <16504129+sagudev@users.noreply.github.com>
This change allows that only new display list is built when css
background and border image loaded.
Testing: This change should not change any behaviors so covered by
existing WPT tests.
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
The value_from_js_object macro exists to paper over differences between
dictionary types that require rooting (via `RootedTraceableBox`) and
those that do not. However, I need to read the macro source every time I
look at the code that uses it because I can never remember what it's
doing. These changes replace the macro with a trait abstraction that is
clearer, and should be a code size win as well.
Testing: Existing WPT tests suffice.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Elaborate documentation of on_history_changed in WebViewDelegate
Testing: Just documentation.
---------
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Euclid Ye <yezhizhenjiakang@gmail.com>
To find scrolling ancestors, we need to walk up the flat tree and only
consider the elements that are in the chain of containing block
ancestors of an element. `scrollParent` now does this so we can use it
to properly implement `scrollIntoView`.
Testing: There are WPT tests for this change.
---------
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
The concept of a "resource header" is not well specced, since it is
unclear what a "resource" is. That said, it most closely matches a
"response" as part of the navigation params.
With this change, we now delay loading the document until either two
things happen:
1. We reached the end of the file
2. We processed 1445 bytes (as defined by spec)
We initially store bytes in the resource header and then after loading
parse the stored bytes. Any subsequent loading will process as before.
Part of #14024
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
While I adding spec comments to the CSP crate, I discovered two issues:
1. We should only use the last sandbox value (WPT test added)
2. We weren't checking for the scripting sandbox flag in document
Also, the autoplay test should have allowed scripts to run, otherwise
the test doesn't run. Since we weren't checking the flag before, the
test ran fine for Servo. However, it wouldn't run for other browsers.
Also realized that an existing test was pointing to a non-existent file
(since it doesn't have `.sub`). Updated that and confirmed that in other
browsers it now properly works (it no longer shows a 404). However,
Servo now fails that test as we don't fire an load event.
Part of #913
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
JavaScript Number does not have Integer type, except for recently added
`BigInt`. That's why we removed `Int` variant from `JSValue` earlier in
#38748. However, the Serde deserialization is strict: when it expects
`u64`, it cannot deserialize "3.0". But when it expects `f64`, it can
still deserialize "3".
Now, we serialize as i64 if the JS Number has no fractional part. This
not only fixes regression, but also improves the case where we have an
integer not representable by i32 and previously would be parsed as f64
again.
Testing: `./mach test-wpt -r
/infrastructure/testdriver/actions/eventOrder.html --headless --product
servodriver` no longer fails when deserializing HTTP request.
Fixes: #39181
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
This implements the Default trait for `AxesOverflow`, to avoid having to
manually set both axes to `Overflow::Visible`.
Also implements the conversion from `&ComputedValues` to `AxesOverflow`,
which is the also used for overflow viewport propagation.
And moves the `is_inline_box()` checkinto the `ignores_overflow` logic.
Testing: Not needed, no change in behavior
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
The servodriver harness requires preference-gated web platform features
to be toggleable at any point in the browsing session's lifetime, rather
than at startup. To support toggling IndexedDB, we need to ensure the
IDB manager thread is always started.
Testing: Verified when running `./mach test-wpt /IndexedDB --headless
--product servodriver`. We don't run servodriver in CI yet.
Fixes: #39175
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Implement getAll and getAllKeys for IDBObjectStore.
Testing: WPT & Unit testing
Fixes: Part of #6963.
---------
Signed-off-by: Ashwin Naren <arihant2math@gmail.com>
fxhash seems to be unmaintained (see
https://github.com/rustsec/advisory-db/issues/2185) so we should move
away from it.
Additionally, the new crate might be slightly faster.
There is still some cases depending on stylo that have the old fxhash
crate.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
Testing: Changes in Hash should really not show any bugs. And
performance should be comparable.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
CookieStore sets up a route with the resource threads to handle async
communication and needs to later unregister itself when the page
unloads. When attempting to do this in the `Drop` of CookieStore we
panic attempting to retrieve the channel via
`self.global().resource_threads()` because global is already null. This
change stores a reference to the core resource thread in the object to
send the unregister on `Drop`.
Testing: manual testing for crash fix, behavior should be unchanged
Signed-off-by: Sebastian C <sebsebmc@gmail.com>
The spec removed all fields from CookieListItem except for name and
value
Testing: WPT tests cover this already.
Signed-off-by: Sebastian C <sebsebmc@gmail.com>
A LayoutBox typically has one LayoutBoxBase, or none in the case of
`LayoutBox::DisplayContents`. However, `LayoutBox::InlineLevel` can
contain multiple inline items, each one with its base. But since things
like the style or the fragment flags should be the same for all of them,
getting the first base is sometimes enough.
Testing: not needed, no change in behavior.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
When encountering such an ancestor, we were returning null instead of
skipping it.
Testing: Adding new subtest for this. And while I'm at it, another one
for the root element, unrelated to this fix.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This includes some small refactoring and some small breaking changes as
listed below. Other than these I tried to keep the functionality exactly
the same but because in the old code the parsing and settings of
preferences was intermingled it was difficult to figure out.
Small Breaking:
- Size and resources-path were unused but appeared in the help.
- soft-fail and hard-fail: Soft-fail flag got removed because it is too
difficult to keep both. The default is now soft-fail and hard-fail can
be enabled.
- The help strings are obviously formatted differently now.
- -V does not work anymore but -v and --version.
Ideally, we want to have the ServoShellPreferences and Preferences be
directly the Argument structure but that needs a bit more discussion
because it would break backwards compatibility with the commandline.
This increases the binary size by ~280kb.
Testing: The testcases are still working but they do not cover much.
I added a unit test for the -p flag because it is the most difficult to
parse in general.
Fixes: This will fix a small number of various parsing misshaps. It will
also show if we are removing an option via unused lint.
Signed-off-by: Narfinger <Narfinger@users.noreply.github.com>
The current implementation of evaluate_key_path_on_value was translated
from gecko, and it is incomplete. The unimplemented part occurs many
crashes in WPT tests.
This PR re-implements it according to the spec. It should eliminate many
crashed WPT tests, and increase the code readability.
Testing: Update WPT test expectation
Fixes: #38817 partially, and #25325
---------
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
This is in preparation of being able to do mime sniffing on the response
data. For that, we first need to introduce separate methods so that we
can decouple them from process_response. In doing so, we introduce a
NavigationParams which mimics what the spec intents. The spec stores
different data (policy container instead of csp list and response
instead of content-type), but it is similar enough.
Part of #14024
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Implements document's active sandboxing flags. These are currently
populated only from CSP-derived sandboxing flags for a new document,
when defined in the CSP.
Testing: 1 new pass, and some new wpt's are added to test points in the
spec where these flags influence behaviour.
Signed-off-by: Shane Handley <shanehandley@fastmail.com>
Reduce the reliance on standalone helper functions for handling JWK
format. Instead, those functionalities are now integrated into the
`JsonWebKey` type generated by script_binding, via the local trait
`JsonWebKeyExt`, for internal use.
The `parse_jwk` function remains for now. It will be removed when once
we refactor `SubtleCrypto::ImportKey` to support a more generic approach
across different cryptographic algorithms.
Testing: Refactoring. Existing WPT tests should suffice.
---------
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
Accept the safe `JSContext` wraper to this function so that it can be
safe. Some callers also become safe as well.
Testing: This does not change behavior and is thus covered by existing
tests.
Fixes: #39129.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
In the wait-for-all algorithm of the IDL promise type, we need to queue
a microtask to perform successSteps given « » if total is 0.
This step was previously implemented in a workaround, which perform
successSteps immediately.
This patch properly queue the microtask, and remove the workaround.
Testing: Refactoring only. Existing tests are enough.
Fixes: #37259
---------
Signed-off-by: Kingsley Yung <kingsley@kkoyung.dev>
There is no need to separate the two kinds of scrolling boxes into
`Element` and `Viewport` more than once. This also eliminates a
potentially panicking `unwrap()`.
Testing: This doesn't change behavior and is thus covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
While we don't perform any sniffing yet, it does now
check all relevant mime types. Before, we would only
check specific versions of specific mime types. However,
the specification is more elaborate, which MimeClassifier
already handles.
There are two new test failures, but they actually fail
in the same way in other browsers. These tests are still
being discussed what the correct behavior should be [1]
Part of #14024
[1]:
https://github.com/whatwg/mimesniff/issues/189#issuecomment-2081559661
---------
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
More small geometry fixes.
Testing: Covered by WPT
(css/geometry/DOMMatrix-invert-preserves-2d.html)
---------
Signed-off-by: lumiscosity <averyrudelphe@gmail.com>
#39112 restricts accessibility for some font functions, exposing some
functions that haven't been used for a long time. Now it is detected by
Lint as dead code.
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
It more logically matches `MediaList`, and it allows us to
call this method with a document. This is required when
parsing Link headers, as they don't have an associated
element, but they do have a document.
Part of #35035
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
This makes it easier to experiment with the impact of experimental
features when browsing around in servoshell. The toggle is global and
causes all webviews to reload with the new preference values.
Testing: Manually tested; no UI testing for servoshell.
Not enabled:
<img width="317" height="82" alt="Screenshot 2025-09-03 at 9 34 30 PM"
src="https://github.com/user-attachments/assets/ca521ad5-ce1b-434e-a0c3-ea1b75d76d53"
/>
Enabled:
<img width="320" height="82" alt="Screenshot 2025-09-03 at 9 34 36 PM"
src="https://github.com/user-attachments/assets/7b6529b5-1055-4ae0-924a-96d57e115714"
/>
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
This change marks all items that are not used by other crates as
`pub(crate)` instead of `pub`.
Additionally, I've removed all code that turned out to be unused. I
think most of it was used only by `layout-2013`.
Testing: The correctness of these changes is verified by the compiler,
so no tests are needed.
---------
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
