Instead of manually triggering `ScriptThread::update_the_rendering`,
have animated images trigger rendering updates via the `ScriptThread`
event loop. This should result in fewer calls to
`ScriptThread::update_the_rendering`.
Testing: This should not change behavior and is thus covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Instead of generating a frame for every display list, which might be one
rendered frame per `<iframe>`, generate only a single frame per call to
"update the rendering." This should make rendering more efficient when
there are `<iframe>`s present and also open up optimizations for
non-display list frames.
Testing: This could potentially reduce flashing of content during
rendering
updates, but that is very difficult to test.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This change makes it so that every `GlobalScope` can contain an optional
`FontContext`. This will be necessary for every `GlobalScope` that
accesses canvas. Currently, `FontContext` is created and accessed via
the canvas worker thread, but this means that web fonts are not
available to canvas. This change will eventually make it possible for
canvas to share web fonts with the `Document` that owns them.
Testing: This should not change behavior and is thus covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
To help ensure our internal threads have shut-down before we deinit
Servo, the last thing the constellation should do, is sending the
`EmbedderMsg::ShutdownComplete`.
Testing: Manual testing by starting Servo and then closing the window.
Fixes: Part of https://github.com/servo/servo/issues/30849
---------
Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
Signed-off-by: Gregory Terzian <2792687+gterzian@users.noreply.github.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
This reverts commit fb1c0a4c48.
Previously in `create_compositor_channel`, the [routing callback][1] was
setup so that a message received on the Compositor's IPC receiver will
be
forwarded to the local receiver using the `CompositorProxy` which also
takes care of waking up the event loop. In #38782, this was changed so
that the routing callbacks simply forwards the message directly without
going via the `CompositorProxy`. This breaks behaviours that rely on the
event loop being woken up on message sending, e.g. updating image frames
for animated gifs.
Since the GenericChannel API doesn't allow custom routing callbacks,
revert this change until we figure out a better solution.
[1]:
d2ccce6052/components/servo/lib.rs (L1114)
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
To ensure a clean-shutdown of Servo, we need to shutdown our internal
threading before deinit. This shuts down the style thread pool either
using the constellation(in single-process mode), or when a content
process exits.
Testing: Manual testing by opening a window, opening and closing
multiple tabs, and closing the window, with a sleep added before deinit
and a sample taken to ensure the loose threads previously noticed are
gone.
Fixes: part of https://github.com/servo/servo/issues/30849
---------
Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
Adds a new memory report that aggregates the size of all scroll trees
from within all pipelines for each web view.
Testing: Acessing `about:memory`
Fixes: #38726
---------
Signed-off-by: criskell <96352451+criskell@users.noreply.github.com>
#38614 was reverted due to CI flakiness, but it also included several
improvements to the devtools tests. this patch relands those
improvements, described below.
we make three changes that speed up the devtools tests from 73 → 65 → 56
→ 51 seconds:
- we replace the hardcoded sleep(1) after starting servoshell with a
loop that waits until the devtools port is open (this also fixes
intermittent failures when servoshell starts too slowly, especially on
macOS)
- we start the internal web servers once, and reuse them across all
tests
- we run servoshell in headless mode (this is also required because most
CI runners have no GUI)
and we fix two bugs that cause very noisy but not very interesting error
messages:
- in the test code, we use a [context
manager](https://docs.python.org/3/reference/datamodel.html#context-managers)
to ensure the devtools client is disconnected unconditionally, even if
test methods or assert helper methods raise exceptions (this was causing
errors on all platforms)
- in the devtools server, we treat “connection reset” errors when
reading from the client like a normal EOF, rather than as a failure
(this was causing errors on Windows)
on self-hosted linux builds, there are still spurious error messages
like the following, but we can fix them later:
```
error: XDG_RUNTIME_DIR not set in the environment.
libEGL warning: egl: failed to create dri2 screen
```
Testing: this patch improves the devtools tests, but does not change
their coverage
Fixes: part of #36325
---------
Signed-off-by: Delan Azabani <dazabani@igalia.com>
Signed-off-by: atbrakhi <atbrakhi@igalia.com>
Co-authored-by: atbrakhi <atbrakhi@igalia.com>
- Add a helper function `focus_and_raise_to_top` to `WebView`.
- For webdriver, raise webview to top when focus, to make it consistent
with user interaction.
Testing: Should reduce some flaky TIMEOUT when combined with another PR
later.
Fixes: A small patch before the PR for #38906.
---------
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
This adds the remaining window as well as specific svg and animation
listeners. The test suite was erroring before, as we don't implement
`SVGAnimationElement` yet. Now, the test gracefully checks if the
interface exists before doing a lookup.
Part of #36258
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Remove the intermediate clone of an IpcSender, by directly sending.
Testing: No functional changes
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
The method now doesn't need unsafe in its signature because it no longer
accepts unsafe pointers as arguments. We move the unsafe marker to the
method itself.
Testing: I opened the browser and went to google.com; I ran some WPT
(IndexedDB) tests.
Fixes: #38361
---------
Signed-off-by: criskell <96352451+criskell@users.noreply.github.com>
The IpcSend trait defines a `send()` method, so doing .sender().send()
seems like it just adds a useless clone of the sender, when we could
just `send()` directly.
This only cleans up the direct usages of this pattern, there are more
instances, where a helper method is defined, which returns the
IpcSender, and the only usages also just directly call send.
Testing: No functional changes
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
Previously, the script-thread would assert the pipeline was closed if no
pending load was found, but it did not check whether the pipeline was
closed before processing the page headers. Since incomplete loads are
removed only when page headers are processed, this means the page
headers were processed even if the pipeline had been closed before the
page headers were available. If the pipeline had been closed as part of
exiting the constellation, it was possible for the constellation to have
exited by the time the page headers became available(since the
script-thread closes a pipeline independently from ongoing navigation
fetches), which would produce a panic on trying to communicate with the
constellation to obtain the browsing context info.
Note: due to the nature of the problem, I cannot verify that this fixes
the crash test, although logically this appears to make sense, and a
couple of days of WPT runs should tell us more.
Testing: A crash test was added; unfortunately the crash was
intermittent.
Fixes: https://github.com/servo/servo/issues/36747
---------
Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
Ports the channel for WebDriverLoadStatus to GenericChannel.
Testing: No functional changes - Covered by existing webdriver tests
Part of #38912
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
Previously, we would always mark the image-related nodes as dirty
whenever the fetch status of the image resources changed. However, the
corresponding `ImageDisplayItem`s for these image resources are only
generated after the image resources have been fully fetched and decoded.
Therefore, we only mark the corresponding DOM nodes as dirty when the
image resources are completely loaded, thereby reducing the occurrence
of reflows.
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
Migrates the namespace sender and receiver to use GenericChannel
Testing: Covered by existing tests
Part of #38912
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
This is a step toward fixing Rust warnings about unsafe code needing to
be wrapped in `unsafe {}` blocks.
Testing: This does not change behavior is thus covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Besides migrating the channel to GenericChannel, this PR adds
`routed_channel_with_local_sender()` to `generic_channel`. This is for
existing use-cases, where we want to provide both an IPC capable
GenericSender, as well as a crossbeam Sender, for efficient sending if
the sender is in the same process.
Testing: All of our channels should send / receive at least some
messages during WPT tests.
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
In Single-process mode, sending crossbeam channels over an ipc channel
is perfectly safe to do (since everything is in the same process) and
allows us to more easily incrementally port channels.
In Multi-process mode, GenericChannels will always be IPC channels, so
we won't hit the "serialize crossbeam channels" branch (since the only
way to construct channels, checks the process mode). This property is
ensured by `channel()` being the only way to construct a `GenericSender`
and Receiver pair. To achieve this, we make the previously `pub` enum
private, and wrap it in a newtype, so that the type can't be constructed
from outside the module.
To be extra safe, we still check if we are in multiprocess mode or not
during (de-)serialization and emit an error.
Testing: Add a new unit-test to ensure sending GenericSender / Receivers
over an ipc_channel works.
---------
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
Move `allow(too_many_arguments)` to the functions that have that issue
and remove `allow(unreachable_pattern)`. Since the `Canvas` enum is also
using conditional compilation the default pattern is not necessary.
Testing: This should not change any behavior and thus is covered by
existing tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
We should not be using `route_ipc_receiver_to_new_crossbeam_receiver` or
similar methods, that `unwrap()` on the ROUTER thread if they encounter
IPC errors. Instead, we now propagate the error to the crossbeam
receiver.
In the GenericChannel::Crossbeam case this means, that we need to use a
`Result<T>` as the data type, even though the Result variant is always
okay, so that the receiver type is the same regardless of `IPC` or not.
This is required, so we have the same channel type, and can pass the
inner crossbeam channel into e.g. `select!`, without having to wrap or
re-implement select.
This also means, that as we switch towards GenericChannel, we will
gradually improve our error handling and eventually remove the existing
panics on IPC errors.
These changes were extracted out of
https://github.com/servo/servo/pull/38782
Testing: Covered by existing tests. No new panics were introduced.
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
This is necessary so that `constellation_traits` can get these types via
a
dependency on `fonts_traits`. This will allow sending IPC channels to
shared
workers so that they can have access to a shared `FontContext` from
`script`.
Testing: This just moves code between crates, so is covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Add logic that was accidentally removed in #38745. Otherwise it is very
weird when using webdriver as a human, as we still stays on the original
tab.
Testing: ~this should not affect any test. Even if the tab is not
"visible" previously, all programmatic interaction works fine.~ Stably
pass `test_history_pushstate` in `back.py` and `forward.py`.
---------
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
Moves interfaces defined by the webaudio spec (27 files) to the new
`script/dom/audio/` module from the `script/dom/` module.
Testing: Just a refactor shouldn't need any testing
Fixes: N/A
Signed-off-by: Ashwin Naren <arihant2math@gmail.com>
The `font-types` crate is the tiny type-only base crate of `fontations`.
This uses a strongly typed representation of open type tags, and allows
us to remove our custom macro for creating them.
---------
Signed-off-by: Nico Burns <nico@nicoburns.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
These changes fix a large number of panics that can manifest as
intermittent test failures. They also add more specification text to
various IDBObjectStore methods and implement missing steps that check
for whether an object store is deleted.
Testing: Existing test coverage.
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
We wouldn't handle the AttributeMutation::Removed for
attribute event listeners and wouldn't remove the
corresponding event listener. Added the necessary
logic (using the newly EventTarget::is_content_event_handler
to correctly only do this for known event handlers) and
added links to the relevant parts of the spec.
Signed-off-by: Tim van der Lippe <tvanderlippe@gmail.com>
Instead of iterating through all frames every time we are looking for a
particular frame, just index into the `Vec` of frames directly. This
should do less work on every frame transition.
Testing: This is just a small optimization, so shouldn't change
observable
behavior.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Testing: Semantically it is the same. But still run the test locally and
in [try](https://github.com/yezhizhen/servo/actions/runs/17159323926) as
I was worried about some unexpected side effect from
176e42d36d. But seems everything is good.
---------
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
Bumps `vello`, `vello_cpu`,`skrifa`, and `read-fonts` dependencies.
Notes:
- The primary motivation for this is so that our version of `read-fonts`
matches the version used by current releases of `harfrust`
- `vello_cpu` now defaults to `RenderMode::OptimizeSpeed` so we don't
need to pass this explicitly.
Signed-off-by: Nico Burns <nico@nicoburns.com>
The new version of the functions return `Result` types.
Testing: No tests because this should not change behavior.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Instead of copying the font table data in memory and parsing it with the
`truetype` crate, use a non-copying API from DirectWrite to implement a
`fontations` `TableProvider`. This has two benefits:
- We remove the dependency on the `truetype` crate finally.
- We do not have to make an in-memory copy of the table data when
parsing the table.
The hope is that the `TableProvider` will be more generally useful in
the future.
Testing: There are no automated tests for Windows, but I manually
verified
that the data retrived via `fontations` matched that retrived by
`truetype`.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This is an implementation of the `prevent_scroll` feature in the focus
transaction system. It allows to control whether focusing an element
should prevent scrolling or not.
Spec:
https://html.spec.whatwg.org/multipage/interaction.html#dom-focusoptions-preventscroll
Testing: Existing WPT tests
Signed-off-by: abdelrahman1234567 <abdelrahman.hossameldin.awadalla@huawei.com>
Unlike other platforms where we read the default axis values and combine
it with variations from style to make the font face, we set the
variations from the style when creating the font face and then read the
final variations from the face. It seems that DirectWrite does the
normalization of variation values internally.
This depends on servo/dwrote-rs#68.
Testing: We currently don't have tests for Windows, but variation
support is
covered by the WPT tests.
Fixes: This is part of #38800.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This is necessary to fully support font variations.
Testing: This will be tested once variations are enabled.
Fixes: This is part of #38800.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
While building the stacking context tree we were assuming that `<body>`
would have propagated its `overflow` value to the viewport, and thus its
used `overflow` would be `visible`.
However, the element that propagates `overflow` can be the root element
instead. Since #38598 we are correctly taking this into account in
`effective_overflow()`, so we no longer need to do anything special in
the stacking context logic.
Testing: `css/css-overflow/overflow-body-propagation-012.html`
Fixes: #38799
Signed-off-by: Shubham Gupta <shubham13297@gmail.com>
This changes makes a variety of changes to ensure that the cursor is set
back to the default cursor when it leaves the `WebView`:
1. Display list updates can come after a mouse leaves the `WebView`, so
when refreshing the cursor after the update, base the updated cursor
on the last hovered location in the `DocumentEventHandler`, rather
than the compositor. This allows us to catch when the last hovered
position is `None` (ie the cursor has left the `WebView`).
2. When handling `MouseLeftViewport` events for the cursor leaving the
entire WebView, properly set the
MouseLeftViewport::focus_moving_to_another_iframe` on the input event
passed to the script thread.
3. When moving out of the `WebView` entirely, explicitly ask the
embedder to set the cursor back to the default.
Testing: This change adds a unit test verifying this behavior.
Fixes: #38710.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Testing: WebDriver Conformance test. In addition to fixing all "Element
Intercepted" errors in
https://github.com/yezhizhen/servo/actions/runs/17142506541, we are able
to fix many other tests.
Fixes: #38837
---------
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
Previously we immediately passed the KeyboardEvent to embedder. Now we
make element send keys go through the dispatch action which required by
spec. CompositionEvent still immediately passed through embedder
Testing: Should make
`./tests/wpt/tests/webdriver/tests/classic/element_send_keys/` more
stable.
Fixes: https://github.com/servo/servo/issues/38354
Fixes: https://github.com/servo/servo/issues/38442
---------
Signed-off-by: PotatoCP <Kenzie.Raditya.Tirtarahardja@huawei.com>
Co-authored-by: Euclid Ye <euclid.ye@huawei.com>
In the `IDBObjectStore::new` constructor, the `transaction` field is
initialized to null, but when using this constructor, we always execute
`set_transaction` immediately afterward. Therefore, we refactored to
require the `transaction` field to be specified during construction and
thereby also removed some no longer necessary assertions.
We also updated the `transaction` field in WebIDL to remove the nullable
capability.
Testing: WPT
Fixes: #38814
---------
Signed-off-by: criskell <96352451+criskell@users.noreply.github.com>
Servo shows a behavior unlike FF and Safari(I don't have Chrome), where
stopping a window does not cancel planned form navigation, resulting in
an infinite navigation loop. The current behavior of Servo does seem to
follow the wording of the spec, so I will open a [companion issue at the
spec](https://github.com/whatwg/html/issues/11562), and I have also
written a WPT tests for the non-standard but widely followed behavior.
This PR also adds a beginning of an implementation of the "ongoing
navigation" concept, which is used by the spec to cancel navigations,
and which is used in this PR only to cancel planned form navigations.
The generation id concept, which corresponds to the planned navigation
concept in the spec, is turned into a simple struct private cell, and is
documented per the spec.
Testing: A new WPT test is added
Fixes: Only one part of https://github.com/servo/servo/issues/36747
---------
Signed-off-by: gterzian <2792687+gterzian@users.noreply.github.com>
Signed-off-by: Gregory Terzian <2792687+gterzian@users.noreply.github.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Following #37527, every textual input is constructing the containers for
value and placeholder. However not all of textual `<input>` element
require the initialization of such placeholder container. This is
apparent with JS UI framework that defines its own placeholder
management.
This PR add lazy initialization for placeholder which construct the
relevant HTML elements for placeholder container whenever it is
necessary.
Testing: Existing WPT coverage
Signed-off-by: Jo Steven Novaryo <jo.steven.novaryo@huawei.com>
We don't need to serialize internal String and then serialize the Result
type.
Testing: Tested. No behaviour change.
Signed-off-by: Euclid Ye <euclid.ye@huawei.com>
#38819 made a step in the right direction. Unfortunately sqlite doesn't
support unsigned integers, so I've been storing them as i64s internally,
but deserializing the bytes to u64s. This allows for an extra bit of
information, but by inserting 0 into the table, it was interpreted
`u64::from_ne_bytes([1,0,0,0....,0])` (or whatever the internal bit
representation of `0_i64` is on the platform), which is not intended.
Signed-off-by: Ashwin Naren <arihant2math@gmail.com>
