This patch should not impact behavior; it is purely a refactoring change.
In a later patch, I will add more (non-optional) members to this new class, to
cache additional data.
One happy outcome of the current patch is that CachedBAxisMeasurement's members
can now become (mostly) 'const' again, now that we can reconstruct it in-place
by virtue of it being stored in a Maybe<>.
Differential Revision: https://phabricator.services.mozilla.com/D67802
--HG--
extra : moz-landing-system : lando
This patch should not impact behavior; it is purely a rename.
The "reflow result" naming is too generic for this struct, now that I'm
planning to cache other results from flex item reflow (in later patches).
This particular struct is caching block-axis measurements (a BSize and an ascent), so let's rename it to reflect that.
Differential Revision: https://phabricator.services.mozilla.com/D67801
--HG--
extra : moz-landing-system : lando
This patch should not impact behavior; it is purely a refactoring change to do the same arithmetic in a different spot.
The content-box size is the size that we actually care about when we use this
cached data, so we might as well just store that size directly, instead of
storing the border-box size and making adjustments at usage time, which is what
we do right now.
Note that one of the usage sites had an informational NS_WARNING_ASSERTION, for
cases where the BorderPadding subtraction might drop us into negative
territory. I'm not particularly concerned with preserving that warning at this
point (in part because it's not necessarily an error when it fails, since it's
possible to make it fail with huge sizes; and in part because it's non-fatal,
which means we're not likely to notice it if it did "legitimately" fail
anyway). So I've just removed it for simplicity & consistency.
Differential Revision: https://phabricator.services.mozilla.com/D67800
--HG--
extra : moz-landing-system : lando
I'm unsure of the root cause as the file should exist if generate_tasks has worked, but this should avoid errors in the meantime while we investigate.
Differential Revision: https://phabricator.services.mozilla.com/D67833
--HG--
extra : moz-landing-system : lando
This follows the IonBuilder code but the WarpBuilder design makes this come out
a bit simpler and nicer.
Also cleans up BytecodeLocation to not conflate jump targets and TableSwitch's
default-target. It worked, but it's an implementation detail and BytecodeLocation
should be explicit about the difference.
Differential Revision: https://phabricator.services.mozilla.com/D68012
--HG--
extra : moz-landing-system : lando
Note that the resumeAt in IonBuilder for JSOp::Debugger made little sense: MDebugger
bails out if an onDebuggerStatement hook is active but that uses the previous resume
point and not the one attached there. Strictly speaking MDebugger isn't effectful
at all so it doesn't need a resume point, but changing getAliasSet is probably not
worth it.
Unfortunately the instrumentation callback can be nursery allocated. For now
WarpOracle aborts compilation in this case, I think that's fine considering the
instrumentation machinery isn't really used anymore.
Finally, maybe in the future WarpOracle should allocate WarpScriptSnapshot upfront
so we don't need a constructor with many parameters.
Differential Revision: https://phabricator.services.mozilla.com/D68004
--HG--
extra : moz-landing-system : lando
The high-contrast light GTK theme has some known issues. The dark theme works
much better, but we have multiple complaints about high contrast users, some of
which don't know about the "override document colors" setting.
Disable it for now so as to go to the previous state for the time being.
Mac/Android don't have a high-contrast theme either (or we don't hook into it),
so 1 seems like a better / less confusing default.
Differential Revision: https://phabricator.services.mozilla.com/D68066
--HG--
extra : moz-landing-system : lando
The bocked.svg used as img in Network Monitor is draggable that
isn't consistent with the rest of the UI. This problem doesn't
exist for the icon used to toggle Network Blocking Panel.
Taking cue from the same implementation, converted the image
to a CSS background property.
Differential Revision: https://phabricator.services.mozilla.com/D67909
--HG--
extra : moz-landing-system : lando
so that references can be released from the main thread and there is no need
to send a message to Console thread.
Depends on D67999
Differential Revision: https://phabricator.services.mozilla.com/D68000
--HG--
extra : moz-landing-system : lando
This provides that ConsoleRunnable no longer has a reference to Console, which
previously needed to be released through a message to the console thread.
Differential Revision: https://phabricator.services.mozilla.com/D67999
--HG--
extra : moz-landing-system : lando
to remove Console instance access from
PopulateConsoleNotificationInTheTargetScope().
Differential Revision: https://phabricator.services.mozilla.com/D67998
--HG--
extra : moz-landing-system : lando
This will support removal of Console class usage from main thread.
Depends on D67993
Differential Revision: https://phabricator.services.mozilla.com/D67994
--HG--
extra : moz-landing-system : lando
Now that ConsoleCallData has nothing to trace, mCallDataStoragePending has no
purpose.
Depends on D67989
Differential Revision: https://phabricator.services.mozilla.com/D67990
--HG--
extra : moz-landing-system : lando
so that all ConsoleCallData members can be destroyed on either thread.
ArgumentData::mArguments hold the same references that
ConsoleCallData::mCopiedArguments held previously. The name change is because
the references are merely stored rather than any deep copy of objects.
Differential Revision: https://phabricator.services.mozilla.com/D67989
--HG--
extra : moz-landing-system : lando
This removes a dependency on JS objects on ConsoleCallData, and a reference to
the arguments on ConsoleProfileWorkerRunnable.
Differential Revision: https://phabricator.services.mozilla.com/D67988
--HG--
extra : moz-landing-system : lando
- Add functionality to clone adopted style sheets for printing.
- Add reftest to ensure that the document's adopted styles show in print.
- Add reftest to ensure that a shadow root's adopted styles show in print.
Differential Revision: https://phabricator.services.mozilla.com/D66517
--HG--
extra : moz-landing-system : lando
By painting focus colors. I suspect this was mostly an oversight? But it is the
most obvious issue I always find with this theme.
I followed active > focus > hover, which seems to match what GTK does (and makes
sense, generally).
Differential Revision: https://phabricator.services.mozilla.com/D68088
--HG--
extra : moz-landing-system : lando
Only button / menulist-button were missing from the hard-coded if condition. I
don't think we ever want to override author padding, and this can cause compat
issues as the one in this bug.
I'm making HasAuthorSpecifiedRules fast in bug 1624080, btw.
Differential Revision: https://phabricator.services.mozilla.com/D68085
--HG--
extra : moz-landing-system : lando
The texture array is currently grown layer by layer and we typically get to 3 or 4 layers over several frames by the time we are done loading a simple wikipedia page.
Differential Revision: https://phabricator.services.mozilla.com/D68056
--HG--
extra : moz-landing-system : lando
The current heuristic in TextureCache::maybe_reclaim_shared_memory pretty much clears the cache every 5 seconds. Clearing the cache is prtty drastic though, because it causes us to re-upload data and reallocate several textures on the next frame. We really only want to do it when the savings are big, which happens less often now that texture array layer count is capped at 16 and that textures are released as soon as they are empty.
This makes us clear the cache less often by augmenting the threshold to 16 megabytes and only considering texture regions that would not be reallocated right away (since we grow some texture arrays more than one region at a time).
Differential Revision: https://phabricator.services.mozilla.com/D68051
--HG--
extra : moz-landing-system : lando
We already have a cooldown from texture cache items being deallocated a certain amount of time and frames after their last use so we can deallocate texture arrays as soon as they are completely empty. We do this at the end of the frame to avoid deallocating and reallocating within the frame. It's better to reclaim texture memory this way than run into maybe_reclaim_shared_memory which will throw away everything and cause new allocations on the next frame.
Differential Revision: https://phabricator.services.mozilla.com/D68050
--HG--
extra : moz-landing-system : lando
time.clock() is deprecated since Python 3.3 and gone in Python 3.8.
Depends on D67773
Differential Revision: https://phabricator.services.mozilla.com/D67774
--HG--
extra : moz-landing-system : lando
Daniel Holbert
jayati
Julian Descottes
Simon Fraser
Sydney
Mirko Brodesser
Jan de Mooij
Emilio Cobos Álvarez
Nicolas Chevobbe
Jeff Gilbert
Brindusan Cristian
Kriyszig
Dan Glastonbury
Karl Tomlinson
Lee Salzman
Dragana Damjanovic
Daniel Huigens
Erik Nordin
Nicolas Silva
Anthony Ramine
Narcis Beleuzu