Skip to content

GitLab

T
tangled
Switch branch/tag
  1. 21 Dec, 2019 8 commits
    • Brandon Wylie's avatar
      Revert "Reparent the current Tab when the theme is changed" · 0ecee86d
      Brandon Wylie authored 
      This reverts commit 24eca4b9.

Reason for revert: Crashing Tot

Original change's description:
> Reparent the current Tab when the theme is changed
>
> * Creating a class which handles reparenting the current tab when the
>   theme is swapped.
> * Hooking that class into ChromeActivity.
> * Adding plumbing to a few ReparentingTask/TabReparentingParams to
>   facilitate the process.
>
> Bug: 944302
> Change-Id: I2801eba979b3acafa02856a9053bc474bf2f581c
> Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1931763
> Reviewed-by: Yusuf Ozuysal <yusufo@chromium.org>
> Commit-Queue: Brandon Wylie <wylieb@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#726136}

TBR=fgorski@chromium.org,yusufo@chromium.org,wylieb@chromium.org

# Not skipping CQ checks because original CL landed > 1 day ago.

Bug: 944302
Change-Id: If032ff5e4cf61f4047e4cf146c8ae6a03a78aa77
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1979398Reviewed-by: default avatarBrandon Wylie <wylieb@chromium.org>
Commit-Queue: Brandon Wylie <wylieb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726996}
      0ecee86d
    • Marijn Kruisselbrink's avatar
      [FileAPI] Change how snapshots are captured by blink::File. · 2e8a62b7
      Marijn Kruisselbrink authored 
      Instead of going through mojom::blink::FileUtilitiesHost and passing a path
from the renderer back to the browser, add a new method to mojom::blink::Blob
to do the same thing. Since the blob already knows what file a blob is
referencing, this both makes things slightly more secure (or at least gets
us closer to a place where we can get rid of things like
ChildProcessSecurityPolicyImpl::CanReadFile) and makes this
work in a couple more cases after a File is for example postMessaged to a
worker or frame in a different process as well.

Additionally this gets us closer to unifying what the renderer and browser
processes consider to be "the" snapshot state of any particular blob. In
a follow-up CL the browser side code will change to return its snapshot
state (if any), rather than always returning the current state on disk.

This is part of a series of changes to improve consistency surrounding how
snapshot state of files/blobs is implemented.

Bug: 844874
Change-Id: Ia05cfd21a3d78955ff70933995e93bbe59e3bb3b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1970301
Commit-Queue: Marijn Kruisselbrink <mek@chromium.org>
Reviewed-by: default avatarTom Sepez <tsepez@chromium.org>
Reviewed-by: default avatarDaniel Murphy <dmurph@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726995}
      2e8a62b7
    • Khushal's avatar
      blink/images: Avoid painting invalid images. · fc67ce7b
      Khushal authored 
      If the frame count of an image decreases after appending more data to
it, the image is assumed to be broken. Avoid painting such images. This
can result in fatal errors in the downstream code.

And also miscellaneous cleanups.

R=pdr@chromium.org

Bug: 867479
Change-Id: I90b5c7030f80076fdc44beb88910667c6182cb8b
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1972904
Commit-Queue: Khushal <khushalsagar@chromium.org>
Auto-Submit: Khushal <khushalsagar@chromium.org>
Reviewed-by: default avatarEric Karl <ericrk@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726994}
      fc67ce7b
    • Nate Fischer's avatar
      AW: fetch flag overrides via ContentProvider · 34df84d5
      Nate Fischer authored 
      This CL completes the basic implementation of the flags UI. During
startup, the embedded WebView implementation will check if the user has
enabled developer mode, and if so, fetch the flag overrides from the
service.

This uses a ContentProvider instead of an aidl method on the service
interface for the sake of a simple synchronous IPC. Although aidl itself
supports synchronous IPC, the Android framework only supports binding to
the service asynchronously. We require fully synchronous IPC so we can
block startup while we fetch the flags.

Now that we've settled on using a ContentProvider to plumb information
from the developer UI to embedded WebViews, this changes "developer
mode" to be defined by the ContentProvider's state rather than the
Service's state, which has the side benefit of simplifying some of the
Activity/Service code.

We rely on PackageManager APIs to check if developer mode is enabled.
The check itself should have very little impact to startup time, since
PackageManager caches its state in RAM. I benchmarked this check (when
developer mode is disabled) at 0ms on my Google Pixel 2 device. For
simplicity, we do not care about performance when developer mode is
enabled, as this is not the usual user experience.

Bug: 981143
Test: Manual - toggle debug border flag, start WebView shell, see borders
Test: Benchmark isDeveloperModeEnabled() with System.currentTimeMillis()
Test: run_android_webview_junit_tests -f *ServiceNamesTest*
Change-Id: I7cc67d1bdf8f0f2ce0fce714fb359160899354a7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1977828
Commit-Queue: Nate Fischer <ntfschr@chromium.org>
Reviewed-by: default avatarRichard Coles <torne@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726993}
      34df84d5
    • Jordan Bayles's avatar
      Roll src/third_party/openscreen/src/ 2c5e3d4c9..8b5e9a483 (4 commits) · 606e6049
      Jordan Bayles authored 
      https://chromium.googlesource.com/openscreen/+log/2c5e3d4c9d60..8b5e9a4830ef

$ git log 2c5e3d4c9..8b5e9a483 --date=short --no-merges --format='%ad %ae %s'
2019-12-20 jophba Fix build issues in Chrome causing unit tests build to fail
2019-12-18 rwkeane mDNS: Add Known Answer Suppression to Querier
2019-12-20 rwkeane mDNS: Responder Known Answer Suppression
2019-12-19 btolsch Clear fqdn in mDNSResponder setup

Created with:
  roll-dep src/third_party/openscreen/src

TBR=mfoltz@chromium.org

Change-Id: Ic815d00d8e490271c395b9caaba56e454471fa25
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1978845Reviewed-by: default avatarYuri Wiitala <miu@chromium.org>
Commit-Queue: Jordan Bayles <jophba@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726992}
      606e6049
    • Balazs Engedy's avatar
      Name Permissions.Action.WithDisposition.* that in code too. · c55e1cfe
      Balazs Engedy authored 
      These histograms were recorded under the incorrect name:
  Permissions.Prompt.Disposition.*

Bug: 986737, 1036520
Change-Id: I6d27725257bbc96d1b18ec20c67d7e28533b6ac4
TBR: andypaicu@chromium.org
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1978692
Commit-Queue: Balazs Engedy <engedy@chromium.org>
Reviewed-by: default avatarBalazs Engedy <engedy@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726991}
      c55e1cfe
    • Chris Harrelson's avatar
      Only invoke text and element fragment anchors after layout. · e6b10ae2
      Chris Harrelson authored 
      Previously, they could be invoked directly by load events (which would
sometimes force layout first, and sometimes not). This is
problematic because layout may not be clean, and forcing layout
is not straightforward or desirable in these cases.

Now the fragment anchors can only be invoked during the main lifecycle
or a forced layout (*), and in both cases after layout is done. In order
to make sure that one of these occurs during page load, queue an
animation frame after finishing load of the document (this is where
previously it would have synchronously invoked the fragment anchor),
or setting a fragment anchor.

(*) We still need fragment anchors to be invoked by forced layouts,
because script is allowed to change anchors and read back their results
synchronously, and this is necessary for web compatibility.

Bug: 1018632

Change-Id: I6121962e12e50913a58a80330a3b5de96fef5f4f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1935508
Commit-Queue: Chris Harrelson <chrishtr@chromium.org>
Reviewed-by: default avatarDavid Bokan <bokan@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726990}
      e6b10ae2
    • Manas Verma's avatar
      [Autofill Auth Refactor] Wrapping all base::Value with base::Optional · 73162207
      Manas Verma authored 
      This will allow to replace ".is_dict()" or ".is_none()" checks to ".has_value()"
checks.

Bug: 949269
Change-Id: I5cfd37515cdbed4f765b80bfc213877a2458d0db
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1972948
Commit-Queue: Jared Saul <jsaul@google.com>
Reviewed-by: default avatarJared Saul <jsaul@google.com>
Reviewed-by: default avatarSiyu An <siyua@chromium.org>
Cr-Commit-Position: refs/heads/master@{#726989}
      73162207
  3. 20 Dec, 2019 32 commits