epoch is stored as double in ICU. Storing it as usize caused overflow on 32-bit Windows 7
which resulted in dates being formatted to epoch=0.
This patch switches to use f64 on the Rust side.
Depends on D66504
Differential Revision: https://phabricator.services.mozilla.com/D66872
--HG--
extra : moz-landing-system : lando
The patch adds similar to mousemove compression on IPC layer.
https://searchfox.org/mozilla-central/rev/49ed791eec93335abfe6c2880f84c324e73e47e6/dom/ipc/PBrowser.ipdl#764-774
Touch events can't be compressed all the time, because each touch event
may contain changes to several touches.
To disable compression in some cases, another IPC message is sent.
The messages have exactly the same arguments, but by changing the message type,
compression is disabled for that message on IPC layer.
The patch tries to take rather conservative approach by disabling compression
in many cases. Compression is enabled only if the same touches are being changed,
or old touches and some new touch are changed (so the changed touches are a superset of the old touches).
The patch does not do anything with PointerEvent.getCoalescedEvents(). That implementation
would be quite a bit different and riskier for the next coming Fenix release.
Differential Revision: https://phabricator.services.mozilla.com/D63261
--HG--
extra : moz-landing-system : lando
We currently account for this memory but we don't always trigger a GC based on it.
Differential Revision: https://phabricator.services.mozilla.com/D66637
--HG--
extra : moz-landing-system : lando
The pre-write barrier protects against moving an object from in front of the 'marking wavefront' to behind it and leaving it reachable but unmarked. It does this by marking the object when it is removed from it initial position (when the pointer is overwritten). Since weak edges are not marked anyway this barrier doesn't make sense (there is no marking wavefront for weak edges). Instead weak edges must have a read barrier if their value escapes, but that is not check by the pre-barrier verifier.
Differential Revision: https://phabricator.services.mozilla.com/D66620
--HG--
extra : moz-landing-system : lando
If the actual input value didn't change, we'd skip updating form validity, but
that's not always correct because GetStepBase() depends on the value attribute.
Differential Revision: https://phabricator.services.mozilla.com/D66836
--HG--
extra : moz-landing-system : lando
This converts `--setenv` into `env` in `try_task_config` at parameter
generation time.
Differential Revision: https://phabricator.services.mozilla.com/D66537
--HG--
extra : moz-landing-system : lando
There are a number of settings that have equivalent expressions in
`try_options` (used for try syntax) and `try_task_config` (used for other try
selectors). Rather than requiring task generation code to understand both
formats, this converts the try syntax specification to `try_task_config` at
parameter generation time.
Differential Revision: https://phabricator.services.mozilla.com/D66536
--HG--
extra : moz-landing-system : lando
nsFocusManager::GetFocusedWindow may return a window even if focus is actually inside an embedded OOP iframe.
Instead, use nsFocusManager::GetFocusedBrowsingContext, which always knows about the current focus across all processes.
If the focused context is in this process, we get its document and return it.
Differential Revision: https://phabricator.services.mozilla.com/D64514
--HG--
extra : moz-landing-system : lando
When we hit an OOP iframe, we must walk into the child document and continue the search from there.
Differential Revision: https://phabricator.services.mozilla.com/D66711
--HG--
extra : moz-landing-system : lando
Accessible::ChildAtPoint can return an Accessible in a descendant document.
Thus, we must return the result PDocAccessible via IPC, not just the id.
Previously, we only returned the id, but we'd fail when we tried to look it up if it belonged to a descendant document.
Differential Revision: https://phabricator.services.mozilla.com/D66710
--HG--
extra : moz-landing-system : lando
Previously, these stopped at an OuterDocAccessible.
Now, if an OuterDocAccessible is encountered, they continue the search in the remote child document.
As part of this, the bulk of the code has been moved into AccessibleOrProxy, since other callers might find this useful in future.
Also, xpcAccessible::ChildAtPoint now works on Windows.
Differential Revision: https://phabricator.services.mozilla.com/D66534
--HG--
extra : moz-landing-system : lando
The babel-eslint parser defaults to assuming all scripts are modules. Although
we're moving in that direction it seems reasonable to set the default as regular
scripts for now. All the places that were previously overriding the parser are
already specifying the sourceType in their eslint configs.
I chose to put the babel config in a file that babel itself won't look to avoid
the risk of this config impacting the use of babel elsewhere in the tree.
Differential Revision: https://phabricator.services.mozilla.com/D66127
--HG--
extra : moz-landing-system : lando
aarushivij
Zibi Braniecki
Philipp Zech
Christoph Kerschbaumer
Brindusan Cristian
Ted Campbell
Itiel
Noemi Erli
marylicious
Lee Salzman
sonakshi
Olli Pettay
Jon Coppeard
Jeff Gilbert
Hiroyuki Ikezoe
Emilio Cobos Álvarez
Narcis Beleuzu
Tom Prince
Razvan Maries
Jason Orendorff
Masayuki Nakano
James Teh
Dave Townsend
Ed Lee
Andreea Pavel