Now that release promotion is using a hook, all the code for non-hook actions
can be removed.
Differential Revision: https://phabricator.services.mozilla.com/D27204
--HG--
extra : moz-landing-system : lando
Check all transform-like properties which may affect the scaling
factors when computing the suitable scale for animations.
Differential Revision: https://phabricator.services.mozilla.com/D19526
--HG--
extra : moz-landing-system : lando
On android wpt reftests, the window handles list is empty, so we can't use it to switch to a window. Not doing this in that case doesn't seem to break anything, and not doing it in that case prevents the IndexError from being logged to the error log.
Differential Revision: https://phabricator.services.mozilla.com/D27177
--HG--
extra : moz-landing-system : lando
By Bug 1529870, the PrepareForUse() is called in WebRenderImageHost::SetCurrentTextureHost(). It works with single buffer mode android SurfaceTexture for WebGL. But it does not work well with video's SurfaceTexture, since multiple TextureHosts are received and a TextureHost might be skipped. The timing of mCompositableCount becomes from 0 to 1 could be used for calling PrepareForUse().
Differential Revision: https://phabricator.services.mozilla.com/D27044
--HG--
extra : moz-landing-system : lando
When mContinuousUpdate is true, UpdateTexImage() needs to be called for each rendering.
Differential Revision: https://phabricator.services.mozilla.com/D27232
--HG--
extra : moz-landing-system : lando
It's easier this way than fixing who knows how many Webextension APIs that have
learned from Desktop that there is no tab #0 and that therefore refuse to work
in our first tab.
We'll also make a similar change to GeckoView's stub implementation of the tab
API because that affects Custom Tabs and PWAs in Fennec for now.
Differential Revision: https://phabricator.services.mozilla.com/D26431
--HG--
extra : moz-landing-system : lando
Before this patch, the WebAuthnManager/U2F destructors would call MaybeReject on
existing transaction promises. Doing this leaks JS objects. If
WebAuthnManager/U2F are being destructed, though, the window is going away, so
it shouldn't be necessary to reject any outstanding promises. This patch just
clears the transactions.
Differential Revision: https://phabricator.services.mozilla.com/D27346
--HG--
extra : moz-landing-system : lando
`Async.jankYielder` is known to, unfortunately, cause jank by creating a lot of
immediately resolved promises that must be then GCed. For a collection of 50
items, it will create 50 promises and 49 of them will immediately resolve.
Instead of `Async.jankYielder`, we now have `Async.yieldState`, which simply
keeps track of whether or not the caller should yield to the event loop. Two
higher level looping constructs are built on top of it:
* `Async.yieldingIterator`, which has been rewritten to not create extraneous
promises; and
* `Async.yieldingForEach`, which is a replacement for awaiting
`Async.jankYielder` in a loop. Instead, it accepts the loop body as a
function.
Each of these can share an instance of an `Async.yieldState`, which allows an
object with multiple loops to yield every N iterations overall, instead of
every N iterations of each loop, which keeps the behaviour of using one
`Async.jankYielders` in multiple places.
Differential Revision: https://phabricator.services.mozilla.com/D26229
--HG--
extra : moz-landing-system : lando
Right now we can end up with Source objects with multiple source actors that
share the same ID, since nothing tries to make them unique. This patch
makes sources created in Jest tests each get an incrementing actor ID.
Differential Revision: https://phabricator.services.mozilla.com/D27134
--HG--
extra : moz-landing-system : lando
With the previous commits done, the 'newSources' actions expose implementation details that we want
to keep private within the file. Nothing outside of these actions should be instantiating Source
objects anymore.
Differential Revision: https://phabricator.services.mozilla.com/D27133
--HG--
extra : moz-landing-system : lando
We move source creation into the action itself so that the action can inspect the current
state of the store in order to decide how to create the source, where previously we were
forced to insert the source and have the reducer itself handle cases where the source
already existed.
Differential Revision: https://phabricator.services.mozilla.com/D27132
--HG--
extra : moz-landing-system : lando
We want the source actions to be responsible for creating the Source objects rather than creating
them ahead of time. This means that the action itself can make decisions about how to handle new
sources appearing, where right now that decision is made too early by the logic queueing the items.
Differential Revision: https://phabricator.services.mozilla.com/D27131
--HG--
extra : moz-landing-system : lando
We want the two types of queue items to have different datatypes, so we clearly distinguish the
two types of sources that are queued.
Differential Revision: https://phabricator.services.mozilla.com/D27130
--HG--
extra : moz-landing-system : lando
Reducers should be able to use actions to tell what is happening to the workers, and
by combining them into one action they are unable to know what is being added and
what is being removed. Splitting them up means reducers have more information.
Differential Revision: https://phabricator.services.mozilla.com/D27128
--HG--
extra : moz-landing-system : lando
The BaselineScript used to contain the bytecode type map but that's now stored
in TypeScript.
Differential Revision: https://phabricator.services.mozilla.com/D27038
--HG--
extra : moz-landing-system : lando
The local rect for border segments is not solely determined by
the widths and/or radius. Instead of determining the max scale
based on those parameters, use the calculated border segment
rects to determine an appropriate max scale factor.
Differential Revision: https://phabricator.services.mozilla.com/D27189
--HG--
extra : moz-landing-system : lando
This will help the Baseline interpreter restore its interpreterICEntry field
without calling into C++.
Depends on D27032
Differential Revision: https://phabricator.services.mozilla.com/D27033
--HG--
extra : moz-landing-system : lando
Introduce `overflow-wrap: break-word` to CSS declarations in Rules view and Changes panel to prevent long strings from causing unwanted horizontal scroll
Differential Revision: https://phabricator.services.mozilla.com/D27120
--HG--
extra : moz-landing-system : lando
I wrote this while looking at bug 1537903, but actually it doesn't help there,
since all the extra time is spent actually computing styles.
I think this is still worth landing it though.
The reasoning for not caring of this case is that we mint an style anyway out of
the blue anyway. There's no point in restyling the whole document.
Differential Revision: https://phabricator.services.mozilla.com/D27124
--HG--
extra : moz-landing-system : lando
- Remove expectation that 'preventScroll.html' fails.
- Use '[NoInterfaceObject] interface' workaround to simulate missing 'mixin' support.
Differential Revision: https://phabricator.services.mozilla.com/D26922
--HG--
extra : moz-landing-system : lando
This is from Bug 1534967, and the code is wrapped with compiler option
`JS_MORE_DETERMINISTIC`, which is only enabled through
'--enable-more-deterministic' in shell configure.
Differential Revision: https://phabricator.services.mozilla.com/D26656
--HG--
extra : moz-landing-system : lando
Depends on D26920
See Bug 1532993#c12 for the analysis. The extra rendering is due to a window resize that
seems to always happen on ASAN but not on other platforms.
Differential Revision: https://phabricator.services.mozilla.com/D27143
--HG--
extra : moz-landing-system : lando
Depends on D26900
Fixes the blank chart and load issues for the old perf panel in about:devtools-toolbox (or type=content)
Differential Revision: https://phabricator.services.mozilla.com/D26919
--HG--
extra : moz-landing-system : lando