In the first phase, we gather LineItems and then when we have enough to
form a line we turn them into Fragments. This will make it possible to
more simply implement `vertical-align` and `text-align: justify` because
we need to measure the different aspects of the candidate line and then
produce a Fragments.
This is a general refactor of the way that inline layout works, so comes
with some progressions. In addition there are some new failures.
New failures:
Some tests are now failing because only the test or reference is getting
proper line height when it wasn't before. These should be fixed in a
followup change that properly calculate line-height in more cases:
- /_mozilla/css/list_style_position_a.html
- /css/CSS2/floats/float-no-content-beside-001.html
- /css/css-content/pseudo-element-inline-box.html
- /css/css-flexbox/flexbox_flex-none-wrappable-content.html
Some tests are now failing because floats are now placed properly, but
are no longer in their inline box stacking contexts. These will be fixed
by a followup change which properly parents them:
- /css/filter-effects/filtered-inline-applies-to-float.html.ini
- /css/css-color/inline-opacity-float-child.html.ini
One test is failing due to floating point precision errors:
- /css/CSS2/floats-clear/floats-141.xht.ini
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* Add initial support for css-text-3 whitespace handling
This adds initial support for whitespace handling from the CSS
specification for Layout 2020. In general, the basics are covered. Since
test output is very sensitive to whitespace handling, this change
incorporates several fixes:
1. Whitespace is collapsed according to the Phase 1 rules of the
specification, though language-specific unbreaking rules are not
handled properly yet.
2. Whitespace is mostly trimmed and positioned according to the Phase 2
rules, but full support for removing whitespace at the end of lines
is pending on a temporary data structure to hold lines under
construction.
3. Completely empty box fragments left over immediately after line
breaks are now trimmed from the fragment tree.
4. This change tries to detect when an inline formatting context
collapses through.
Fixes#29994.
Co-authored-by: Mukilan Thiyagarajan <me@mukilan.in>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
* Update test results
---------
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Mukilan Thiyagarajan <me@mukilan.in>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
CSS viewport removal
It was removed from the spec and it's disabled everywhere.
This also removes the meta viewport support (which was implemented on top), but that also had a single test and is disabled everywhere, so I'm not too concerned, it can be implemented again if / when needed.
- [x] `./mach build -d` does not report any errors
- [x] `./mach test-tidy` does not report any errors
- [x] These changes fix#26349
- [x] There are tests for these changes
This also removes the meta viewport support (which was implemented on top), but that also had a single test and is disabled everywhere, so I'm not too concerned, it can be implemented again if / when needed.
Previously, final float positions were calculated when their parents
were positioned. This prevented proper positioning of absolute children
of floats with static insets, because they accumulate offsets as they
are hoisted up the tree.
This change moves the final float positioning to
`PlacementState::place_fragment` for the float itself so that it happens
before any insets are updated for hoisted descendants. In addition to
simplifying the code, this makes it a bit more efficient. Finally,
floats are taken into account when updating static insets of hoisted
boxes.
Fixes#29826.
eca0acf459 uncovered a bug in the way that
the scrolling area of `window` was calculated and broke scrolling on the
root element. This change does two things in order to fix that:
1. Does a partial revert of eca0acf459 in
order to get scrolling from script working again on the window
object.
2. Has the compositor always generate a frame for scrolls starting from
script and waits for them. This is speculative fix for flakiness in
root scrolling tests on CI.
This also removes imported position: sticky tests from the Mozilla
directory. These were only supposed to be temporary until the upstream
version from WPT were imported.
Fixes the test case `/_mozilla/css/css-transition-cancel-event
.html`, which was failing under a specific circumstance.
The observed sequence of events during the failing test run looks like
this:
1. Transitions start in `div1` and `div2`.
2. `div1` generates a `transitionend` event.
3. The `transitionend` event handler removes `div2` from DOM, cancelling
its ongoing transition.
4. `div2` is supposed to generate a `transitioncancel` event in a timely
manner, which it does not. The test fails as a result.
What is going on here? Here's a possible explaination:
1. During one invocation of `ScriptThread::handle_msgs`...
2. In step 2, `ScriptThread::update_animations_send_events` -> `Document
::update_for_new_timeline_value` detects the completion of the
transition, and in response, pends the `transitionend` event.
3. In step 3, `ScriptThread::update_animations_send_events` ->
`Animations::send_pending_events` calls the `transitionend` handler.
4. The `transitionend` event handler removes `div2`, thereby cancelling
its ongoing transition and triggering a reflow.
5. Reflow takes place. During this, `Animations::do_post_reflow_update`
-> `Animations::handle_canceled_animations` pends the
`transitioncancel` event (precursor to step 4).
6. Having discovering that there was no running animation, `Animations::
do_post_reflow_update` calls `self.update_running_animation_presence
(_, false)`, which sends `AnimationState::NoAnimationsPresent`.
7. The invocation of `ScriptThread::handle_msgs` ends, and another
starts. It blocks waiting for events.
8. Meanwhile, the compositor receives `AnimationState::
NoAnimationsPresent` and stops further generation of animation ticks.
9. With no events to wake it up, the script thread is stuck waiting
despite having the pending `transitioncancel` event (step 4).
The HTML specification [says][1] that "an event loop must continually
run [...] as long as it exists" and does not say it can block if there
is nothing to do. Blocking is merely optimization in a user agent
implementation. Pending animation-related events must be processed every
time a "rendering opportunity" arises unless the user agent has a reason
to believe that it "would have no visible effect".
Skipping the processing of animation-related events would have visible
effect if such events are indeed present. The correct implementation in
Servo, therefore, would be to request more animation ticks so that such
events are processed in a subsequent tick.
[1]: https://html.spec.whatwg.org/multipage/#event-loop-processing-model
