| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
| |
This is needed by the implementation of loadVisuallyCommitted in the QtWebEngine.
Change-Id: Ic0fcdc00f7963ea1088a1ae910bb2dbe26e9e12c
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
|
| |
Essentially cutting away the unnecessary part that depends on
extensions.
Change-Id: I8adbc77dca393ab8480e9cdea3c5c953b52c8aef
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
| |
As we'll be using a few more files from the chrome layer.
Change-Id: I8008b44d0dba42d7531e6c8fbc0ec050ecca736a
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
|
| |
Removes the addition of 'component=share_library' from GYP_DEFINES.
This caused a build error due to missing files.
Change-Id: I062e7f8ba7d28f298df4bee289e24b495875dc5f
Reviewed-by: Frederik Gladhorn <frederik.gladhorn@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
gcc 4.9
WebGL crashes when chromium is compiled with gcc 4.9,
because TSymbolTableLevel::~TSymbolTableLevel does
double delete of e.g. TFunctions because they are
inserted in the table with both their name and
mangled name.
The fix for this should reach stable with m37.
https://code.google.com/p/angleproject/issues/detail?id=651
Patch by Mike Hommey <mh+mozilla@glandium.org>
https://bugzilla.mozilla.org/show_bug.cgi?id=1025576
Change-Id: I0571478759666bcb55633dd4e630c5b42216d949
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
| |
This is part of using the ANGLE that is built / installed by Qt.
Change-Id: Ib71d47392690dab6cf874aec13db9e98710de684
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This moves libEGL and libGLESv2 from dependencies to
libraries where needed.
This also adds support for using the correct configuration
of the ANGLE libraries for linking and runtime loading.
Change-Id: I8c3a5eb11517bf03e921b257eeb3a34b7bc841ea
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Chromium currently assumes that all GLContexts are checking the availability
of robustness on the system through the same code path. Since we're passing
it a context created by Qt, and that Qt never uses
EGL_EXT_create_context_robustness, the context creation fails when trying to
share a non-robustness-enabled context with one that does.
Fix the issue by asking the shared GLContext if it was allocated using the
extension, Chromium's contexts currently always return true and the Qt context
will return false. This relies on the fact that the Qt context is always the
first one added to the ShareGroup and that it is going to be the one queried,
but this should be reasonably reliable.
Change-Id: I6456429db678f077cd8a72032711361da01e72ea
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes the ASSERT that was triggerd on Windows in
PlatformKeyboardEvent::disambiguateKeyDownEvent because
we do our own conversion from native events to web events.
This is the same way it was handled in the Qt specific
PlatformKeyboardEvent in WebKit.
Change-Id: I177c238bde1e0101b6abe9d5bbe3ba2e47908f1a
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
| |
BUG=72633,337116
Review URL: https://codereview.chromium.org/99473012
Change-Id: I70c2a17ec8a69542c850089b09e19c5873470a1a
Reviewed-by: Michael Bruning <michael.bruning@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
on Windows
BUG=72633
Review URL: https://codereview.chromium.org/111723007
Change-Id: If07088b902b8039ee109e744e46199fd9d13dc46
Reviewed-by: Michael Bruning <michael.bruning@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The linux kernel now reports the processor information
in "model name" and not in "Processor" anymore.
Both should be checked for backwards compatibility.
As described by:
https://code.google.com/p/v8/issues/detail?id=3112#c16
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Change-Id: I567a72c18f43a483e7576e880c2d383873f5660f
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
Since we're never going to statically link to the runtime library on
Windows (unlink Chromium when built as a monolitic chrome.exe), make
sure that our /M flag setting is also properly set for V8.
Reviewed-by: Michael Bruning <michael.bruning@digia.com>
Change-Id: I755c3c0bfa27e45b897a2d4d5496103913d1c6c4
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When replacing substrings with values from a dictionary,
the replacements should not be truncated to the length of src_root_dir.
If the basename does not exist as a key in the resource_ids file,
then fall back to an absolute path.
The precedence is like:
file in chromium > file basename > absolute file name.
THIS SHOULD BE UPSTREAMABLE.
Change-Id: I7d1da8c04515807000742aac4a6fa4db7616fd13
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This adds a let_qmake_do_the_linking target variable to tell ninja
not to link but rather dump the linking information from gyp in a
file that we'll parse with qmake.
Reviewed-by: Zoltan Arvai <zarvai@inf.u-szeged.hu>
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Change-Id: I753888776679e449d7a64b6bf5b47e7a60dfa827
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
| |
Change-Id: Ifbe063950d14d8388e3c8c1d6cebdfae5cbdb8ae
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
| |
Change-Id: I3fcbf166978cf20f7a07d7e6bda135864054e9c1
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
| |
We require that the icu libs are in the default library seach path.
Change-Id: Ia4981781f4239617831909da1250a61191a8c7dd
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
But fall back to chacha_vec_arm.S when arm_neon_optional
is used.
Reviewed-by: Pierre Rossi <pierre.rossi@gmail.com>
Change-Id: I1ea1f03d687c4b39fe4a322abb8ddb91113a1828
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
| |
specific actions
Change-Id: I1bcc9cbb77b10e9e60d60900b9cfd8ce283947b0
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is essentially the last user of the legacy 2d path.
BUG=271140
Review URL: https://codereview.chromium.org/171343003
git-svn-id: svn://svn.chromium.org/blink/trunk@167552 bbb929c8-8fbe-4397-9dbb-9b2b20218538
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Change-Id: I131a9b5ea7487d093c716b9d50933d1ff609c1a9
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
BUG=271575
Review URL: https://codereview.chromium.org/132173005
git-svn-id: svn://svn.chromium.org/blink/trunk@165977 bbb929c8-8fbe-4397-9dbb-9b2b20218538
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Change-Id: Ic26c7821ba5bb5c377c324423d63090ed2f31ffe
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
| |
Rubber-banding is the scrolling effect on Mac when overscrolling
which does not work with Qt and also causes a crash with branch 1750.
Disable the feature for now.
Change-Id: I89f315c2dbb1ed4362b96201566acc784e4bdf96
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
| |
This is needed since https://codereview.chromium.org/106503003
and because of our chromium patch that fixes the build with a
GL ES2 configured Qt.
Change-Id: I2dfa1afe1bb58227656fee798a5d2549f88c3f9d
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
| |
Change-Id: I9997c3dead0f7d37c2a3c45082d5114d0f1632cf
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
| |
Change-Id: I4a494bcda0e963430ab997664728d2a96f72188f
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These threads are using the UI message loop on Windows per default. This
won't work since the UI message loop is handled by Qt and the Renderer
and GPU threads won't work with it properly.
Force these threads for using the default message loop as they use it on
Linux platform.
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Change-Id: I8b04ba9070912378cf1d13173ddff48e98f9a49b
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Add Raspberry PI chipset to list of defined platforms.
- Make sure to include sources for xdg support.
- Do not use pkg-config-wrapper, as our sysroot pkg-config
already returns absolute paths.
- Use openssl for eLinux.
- Use fontconfig and freetype2 for eLinux.
Reviewed-by: Pierre Rossi <pierre.rossi@gmail.com>
Change-Id: Ic5266f818432b4432ed1d8a1ac07f434cb5b86e6
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
| |
Instead we export _override symbols and add the symbol
renaming directly into the QtWebEngineProcess source.
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Change-Id: I15049609dd77be6de2a2398f25a37fff1b6f5519
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
| |
We use RenderWidgetHostViewQt, not RWHVMac, even on Mac.
Since some platform decisions seem to have transpired all the way to
the content layer, we should simply use the web ui for select elements
on mac for the time being.
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Change-Id: Ib430f8686862cf0958ccdbfefb8bfa36af590e97
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes the build with a cross compiler toolchain.
We do not require the functionality of this script but
gyp tries to run it and fails if the specified file
does not exist.
As the upstream issue mentions, this script is "going
away soonish", until then we can keep this patch around.
Reviewed-by: Pierre Rossi <pierre.rossi@gmail.com>
Change-Id: I5de88fcaf750bad4eec2d72a134a86f75a30453e
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
| |
This also removes the aura dependency to chrome resources
since we repack the needed resources ourselves.
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
Change-Id: I0368bb36948ba9612040ed869ef818d4b8471b9b
|
|
|
|
|
|
|
|
| |
platforms
Reviewed-by: Simon Hausmann <simon.hausmann@digia.com>
Change-Id: I5fecc09e245b75d0fec16756f8d6e6b8f71a143e
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
| |
We are currently not planing to enable the sandbox and the error
pollutes the auto tests output.
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Change-Id: I2675aa2e03394f1c03e4df6bd1b21764ed19206a
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Even though plugins are disabled it happens that the libraries are
loaded and cause a crash since skypebuttons pulls Qt4 with it on
Linux and creates a dynamic linking error conflicting with Qt5
symbols.
Add the plugin library name to the blacklist to avoid it being
loaded altogether.
We have a similar fix in QtWebKit doing the same thing.
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Change-Id: I45abf31c4e94a8b61aff85765a91436dabae1b2e
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
| |
This allows Qt-specific IPC messages to be declared in QtWebEngine.
Reviewed-by: Pierre Rossi <pierre.rossi@gmail.com>
Change-Id: I1b401c9c05efc8339c5cacf50f087890447b68a6
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
| |
This will allow us to know right before the first GL context is
instantiated by Chromium so that we can install those contexts to
be shared with QtQuick GL contexts as well.
Change-Id: Id55337a7b42d25ac084698a23dbfab06cb273505
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
| |
This is needed to fetch the MessageLoop, the MailboxManager and the
SyncPointManager of the GPU in-process host.
Change-Id: I7f38e32b2df11da5b046f16643841d34260c11fb
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
| |
This should replace patch 0011 since we do not build GTK code any more.
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
Change-Id: I08ba240fd24e2bd644229458a316fb5886dbd82b
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
GLES/gl2.h is included through Qt public headers and the copy of
Chromium is used since its include path comes before /usr/include.
The problem is that this header is incompatible for some reasons,
one of them being that it converts all GL function symbols from
gl* to GLES2*.
Qt layer code should always need to go through GL directly,
so make sure that only GYP targets that depend directly on gpu.gyp,
khronos.gyp or webkit_gpu.gyp will have an include path
pointing to those headers.
Replace all_dependent_settings with direct_dependent_settings and
control which target inherits this include_dirs from its dependencies
by using export_dependent_settings.
Change-Id: I82ae8a5a62f7d968375b971757b2126670a02461
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
resource_provider.h
If resource_provider.h is going to be used to integrate delegated frame
to external compositors, it should avoid conflicting with the system's gl.h.
Change-Id: Ie37d6cde263157e6717cc006a0896f0b9ebdaa5d
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
| |
Also guard the use of MessagePumpAuraX11 in events_x.cc. We need to
build it to get symbols depending on base::NativeEvent.
Change-Id: Ib7bb4fb1fc8692df1b8151c3d159de3d0f73076f
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
| |
We currently only have a single process binary for Qt.
Therefore we should not choose a different binary based on required privileges.
However, this patch can only be a preliminary solution.
May be we should instead implement our own ChildProcessHost.
Change-Id: I03af6257c7be04ab09c9828d707d30430961a17c
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Add a missing include file to process_iterator_mac.cc.
- Disable Chromium clang plugins.
- Add a use_libcpp variable to common.gypi that can be used
to build with libc++ instead of libstdc++.
In this case several warnings also must be disabled.
- ext/hash_map and ext/hash_set are deprecated.
Use unordered containers as a replacement.
Change-Id: Ief9e8901d0463f2bb62a5960cde03543b3999be1
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
| |
Change-Id: I2bb2600637ac5a3473273789cd61c257d682b702
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This is meant as a baseline commit hence it does not
include the patches we need to apply for QtWebEngine.
All patches should be rebased on top of this commit
so we can get rid of the external patches directory.
In future these baseline commits always have to include
the exact Chromium version returned by version_resolver.py's
currentVersion() in their first line, so that we can
retrieve the patches on top to apply on the upstream
repository. This also includes a ninja update.
Change-Id: I60abeadb785a3b7d149c58b65ddb5a823fed3083
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The linux kernel now reports the processor information
in "model name" and not in "Processor" anymore.
Both should be checked for backwards compatibility.
As described by:
https://code.google.com/p/v8/issues/detail?id=3112#c16
Change-Id: I8828c1a567281e0d7000106338a2033627e8e397
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
These threads are using the UI message loop on Windows per default. This
won't work since the UI message loop is handled by Qt and the Renderer
and GPU threads won't work with it properly.
Force these threads for using the default message loop as they use it on
Linux platform.
Change-Id: Ie3582583e5d1644faa2875e015a0908ba148d91f
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
Reviewed-by: Jocelyn Turcotte <jocelyn.turcotte@digia.com>
|
|
|
|
|
|
|
|
|
| |
The file thread won't do anyting if it uses the Qt event loop therefore
it won't quit properly and this behavior leads to a crash.
Thus the chromium's IO message loop should be used on Windows too.
Change-Id: Ieea0ceb8bd732ed6c9ca172620511d78b0a5d95c
Reviewed-by: Andras Becsi <andras.becsi@digia.com>
|
|
|
|
|
|
|
| |
Gyp needs pkg-config-wrapper now also on eLinux.
Change-Id: Ib4dc7661cde6cd28d156b9a23845aa36f772b55e
Reviewed-by: Zeno Albisser <zeno.albisser@digia.com>
|