There was a typo in the preference name used for the primary list of all regions where VPN promos were allowed to be shown. As a result, VPN promos were shown to regions from a built-in fallback list, which was smaller. This patch fixes the typo and adds a unit test to ensure the code gets its regions from the correct pref.
Differential Revision: https://phabricator.services.mozilla.com/D190628
We are working on [enabling Tier 2 a11y-checks to ensure our products have basic accessibility built-in (bug 1692110)](https://bugzilla.mozilla.org/show_bug.cgi?id=1692110) and before that, we need to prepare the existing code base. While we triage and investigate existent tests that would fail once the a11y-checks are enabled in the CI, we are going to expect to be failing these tests altogether and then handle each one of them individually.
This task is to expect failing of these tests with `fail-if` condition of the failing tests while we continue investigation into the reasons these tests failed and to be able to backtrack these tests once they're resolved.
Related bugs:
- Fail-if tests: meta bug 1848402
- Explore capturing click events for a11y_checks: bug 1692110
Differential Revision: https://phabricator.services.mozilla.com/D190524
This gives a significant windows sp3 improvement, without negatively
affecting memory usage much (on a preallocated content process I get
0.13mb of atoms both with and without this patch).
We could consider bumping it further to 1024, though the relative
improvement vs. this patch is small.
Differential Revision: https://phabricator.services.mozilla.com/D190692
Our current algorithm gives a lot of test failures and test-verfiy failures.
After some logging and debugging, my best guess is that we are just getting
unlucky.
I would see test failures (failed to detect a randomization was applied) when
we e.g. fliped a mere 16 bytes from 0 -> 1. When we flipped some values from
0 -> 2 that would more reliably cause a change.
Differential Revision: https://phabricator.services.mozilla.com/D190283
This is needed for Ubuntu Linux when Firefox is installed as a Snap package.
Under those conditions, Firefox doesn't have the ability to read the contents
of the other browsers' data directories.
We work around by showing the user a native file picker and instructing them
to browse to that data directory in folder picker mode - doing so gives
the Firefox process access to the folder (through a temporary symlink).
Another restriction we must overcome is Sqlite's inability to read from
Sqlite databases through that symlinked folder. We work around this by
copying the databases we're importing from to the system temporary
directory, querying it, and then deleting those copies.
Differential Revision: https://phabricator.services.mozilla.com/D188874
The only noticeable difference is that macOS on dark HCM looks nicer,
effectively.
The hard-coded hex color for the chrome-content-separator is no longer
light there (and in light mode threedshadow is #e0e0e0), and the toolbar
looks better.
Depends on D190685
Differential Revision: https://phabricator.services.mozilla.com/D190686
For some reason these are crashing in beta simulation with `InternalError: too much recursion`.
Instead of trying to pick out the precise set of platforms to skip these tests on, I've just wrapped the guilty code in a try-catch. I tested this locally by increasing the size of the regexp enough to make the test throw on x64 as well.
Differential Revision: https://phabricator.services.mozilla.com/D190622
This fixes the rendering on macOS high contrast. Also removes some now
useless Linux declarations (since they have the same value as the HCM
ones).
Differential Revision: https://phabricator.services.mozilla.com/D190685
The drag service is never instantiated if it's in the headless mode.
And in this case, `NS_ERROR_XPC_GS_RETURNED_FAILURE` exception is thrown.
Therefore, we need to wrap the service getter with a `try` block and
return `null` if it fails.
Differential Revision: https://phabricator.services.mozilla.com/D190580
Call to CSS.registerProperty may result in rendering changes compared
to when variables were interpreted as unregistered custom properties.
Ensure we recascade style for the whole document tree.
This was already covered by WPTs but add a more direct check for
paint and reflow.
Differential Revision: https://phabricator.services.mozilla.com/D190656
Currently wasm ignores GCs triggered by allocation, which can lead to unbounded
memory growth when running purely in wasm. Making the GC interrupt also
interrupt wasm fixes this.
Differential Revision: https://phabricator.services.mozilla.com/D190669
Our arrow is actually more similar to the modern macOS native ones, so
this both looks better and allows us to clean up more code.
flex-end alignment on acceltext-container is not needed (the flex item
is sized to its contents), removing it doesn't change behavior on e.g.
the menubar menus on Linux.
This makes the arrow a bit bigger visually on Linux, but that's what
native GTK does too (see gtk4-demo / gtk3-demo).
Differential Revision: https://phabricator.services.mozilla.com/D190663
This restores the previous behavior though I'm not convinced this code
is useful at all because macOS uses the native menubar instead.
MANUAL PUSH: Orange fix CLOSED TREE
It's a weak pointer but presumably can only die based on main thread actions. The GetBrowsingContext getter is just a plain getter, returning a pointer.
Although small this does show up in the profile in bug 1857094. It's called from some common layout/painting functions.
Differential Revision: https://phabricator.services.mozilla.com/D190559
This patch introduces another timeout to ensure that observers are properly
cleaned up even for pages where the "load" event never happens.
Timeouts are scheduled on DOMContentLoaded and on load. Their duration can
be controlled via the 'cookiebanners.bannerClicking.timeout*' prefs.
Differential Revision: https://phabricator.services.mozilla.com/D190319
Use the same system colors there rather than bespoke ones. That allows
us to simplify the CSS, remove macOS-specific system colors, and also
fix this bug.
Differential Revision: https://phabricator.services.mozilla.com/D190460
Now that we have a DocumentState type we can be a bit less explicit
(before this used EventStates, so the extra Document in the names was
useful).
Differential Revision: https://phabricator.services.mozilla.com/D190602
I think ReleaseAcquire semantics are sound here.
It might be that relaxed is fine too, but that might be technically
unsound because it could make the writes in malloc_init_hard not visible
to the other threads in time.
Depends on D190623
Differential Revision: https://phabricator.services.mozilla.com/D190624
az
Julien Cristau
John Schanck
Dan Mosedale
Heitor Neiva
Paul Adenot
Sandor Molnar
Anna Yeddi
Hanna Jones
Emilio Cobos Álvarez
Andrew Halberstadt
Tom Ritter
Mike Conley
Joel Maher
Paul Zuehlcke
Iain Ireland
Kagami Sascha Rosylight
Dão Gottwald
Timothy Nikkel
Masayuki Nakano
Frederic Wang
Jon Coppeard
Johannes J. Schmidt
Katherine Patenio
Alexandre Poirot
Mirko Brodesser
stransky