| Commit message (Collapse) | Author | Age |
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
https://trac.torproject.org/projects/tor/ticket/19001#comment:11
https://trac.torproject.org/projects/tor/ticket/19001#comment:21
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Build libwebrtc using libcxx. Set use_custom_libcxx=0. According to the
source comment, use_custom_libcxx=1 means to use a precompiled libcxx in
the third_party directory.
Build the test target
obj/webrtc/api/test/peerconnection_unittests.fakeaudiocapturemodule.o
rather than peerconnection_unittests.isolated; building just an object
file doesn't require the linker and is faster.
Use x86_64-apple-darwin11-ar, not ar, to link libwebrtc-magic.a.
When linking go-webrtc later on, it would cause the error:
ld: warning: ignoring file lib/libwebrtc-darwin-amd64-magic.a, file was built for archive which is not the architecture being linked (x86_64)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Descriptions for each patch are in the mbox-formatted patch file.
Subject: [PATCH 01/13] Use flags for GNU grep, not BSD grep.
Subject: [PATCH 02/13] Remove flags not supported by Mac ld.
Subject: [PATCH 03/13] Undefine __APPLE__ in usrsctp.
Subject: [PATCH 04/13] Include <sys/socket.h> in macifaddrs_converter.cc.
Subject: [PATCH 05/13] Allow non-mac to build .m and .mm sources (objc and
Subject: [PATCH 06/13] Propagate "-framework Foundations" in ldflags.
Subject: [PATCH 07/13] Provide definitions of NS_ENUM and NS_OPTIONS.
Subject: [PATCH 08/13] Remove lightweight-generics annotations from NSArray
Subject: [PATCH 09/13] Use -fobjc-arc in webrtc/build/objc_common.gypi.
Subject: [PATCH 10/13] Replace firstObject with objectAtIndex:0.
Subject: [PATCH 11/13] fobjc-arc
Subject: [PATCH 12/13] Use objectAtIndex instead of array index syntax.
Subject: [PATCH 13/13] Hack out some more NS_ENUM and NS_OPTIONS.
|
| |
|
|
|
|
|
| |
As in ab1e5cffbf548f9f586b81928dc22ae90bdec374 for linux.
Update the "$(SDKROOT)" replacement command. Use our own $SDKROOT as the
replacement. Files are now found in additional subdirectories.
|
| | |
|
| |
|
|
|
|
|
| |
The webrtc source extensively uses symlinks, which the "tar -h" in
copy-to-target flattens into duplicate files. Also, the tarball
(omitting .git history) is way smaller than the full source tree (1.5 GB
compared to 13 GB).
|
| |
|
|
| |
This is needed for cross-compilation of libwebrtc for mac.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
http://public.kitware.com/Bug/view.php?id=14603
http://public.kitware.com/Bug/file_download.php?file_id=4981&type=bug
The first public release of cmake that has this fix is v3.0.0.
Without this, setting CMAKE_SYSTEM_NAME=Darwin results in errors:
-- The CXX compiler identification is Clang 3.8.0
-- The C compiler identification is Clang 3.8.0
CMake Error at /home/debian/install/cmake/share/cmake-2.8/Modules/Platform/Darwin.cmake:344 (list):
list sub-command REMOVE_DUPLICATES requires list to be present.
Call Stack (most recent call first):
/home/debian/install/cmake/share/cmake-2.8/Modules/CMakeSystemSpecificInformation.cmake:36 (include)
CMakeLists.txt:15 (project)
-- Check for working CXX compiler: /home/debian/install/clang/bin/clang++
CMake Error at /home/debian/install/cmake/share/cmake-2.8/Modules/Platform/Darwin.cmake:344 (list):
list sub-command REMOVE_DUPLICATES requires list to be present.
Call Stack (most recent call first):
/home/debian/install/cmake/share/cmake-2.8/Modules/CMakeSystemSpecificInformation.cmake:36 (include)
CMakeLists.txt:2 (PROJECT)
CMake Error: Internal CMake error, TryCompile configure of cmake failed
-- Check for working CXX compiler: /home/debian/install/clang/bin/clang++ -- broken
CMake Error at /home/debian/install/cmake/share/cmake-2.8/Modules/CMakeTestCXXCompiler.cmake:54 (message):
The C++ compiler "/home/debian/install/clang/bin/clang++" is not able to
compile a simple test program.
It fails with the following output:
CMake will not be able to correctly generate this project.
Call Stack (most recent call first):
CMakeLists.txt:15 (project)
-- Configuring incomplete, errors occurred!
See also "/home/debian/build/libcxx/build/CMakeFiles/CMakeOutput.log".
See also "/home/debian/build/libcxx/build/CMakeFiles/CMakeError.log".
|
| | |
|
| |
|
|
|
| |
This is a dependency of snowflake since
https://gitweb.torproject.org/pluggable-transports/snowflake.git/commit/?id=e93c38f83473476c8dddf9e107ce4df5bfea7882
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Set to f336982 as in https://github.com/keroserene/go-webrtc/issues/39.
Actually, at this moment, branch-heads/52 is 3 commits ahead (ca7fe7e)
of where it was at the time of GitHub #39.
I had to change "gyp_webrtc" to "gyp_webrtc.py" as the former is no
longer executable:
+ webrtc/build/gyp_webrtc
bash: line 288: webrtc/build/gyp_webrtc: Permission denied
"gyp_webrtc" just does:
execfile(__file__ + '.py')
I added "werror=" to GYP_DEFINES because of this warning that stopped
the build by being treated as an error:
../../webrtc/voice_engine/channel.cc: In member function 'void webrtc::voe::Channel::Demultiplex(const int16_t*, int, size_t, size_t)':
../../webrtc/voice_engine/channel.cc:2954:13: error: '*((void*)& codec +44)' may be used uninitialized in this function [-Werror=maybe-uninitialized]
CodecInst codec;
^
../../webrtc/voice_engine/channel.cc:2954:13: error: '*((void*)& codec +36)' may be used uninitialized in this function [-Werror=maybe-uninitialized]
cc1plus: all warnings being treated as errors
I suspect that more recent versions of GCC than our current 5.1.0 don't
raise a warning here.
I changed the test target name from
libjingle_peerconnection_unittest.isolated to
peerconnection_unittests.isolated. It seems that libjingle is going to
stop existing as a separate concern. Anyway, the old target name doesn't
exist:
+ ninja -C out/Release libjingle_peerconnection_unittest.isolated
ninja: Entering directory `out/Release'
ninja: error: unknown target 'libjingle_peerconnection_unittest.isolated'
|
| | |
|
| | |
|
| |
|
|
| |
Might be used implicitly by other commands, not sure.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
We avoid some dependencies by not building tests. But one of the tests
is needed because go-webrtc uses the FakeAudioCaptureModule class it
defines. So run ninja twice, the first time with tests disabled to build
the library, and the second time with tests enabled to build just the
libjingle_peerconnection_unittest.isolated target.
|
| | |
|
| |
|
|
|
| |
So we can have a copy of it.
https://github.com/keroserene/go-webrtc/issues/38
|
| |
|
|
|
|
|
|
| |
depot_tools
go-webrtc
snowflake
Leaving webrtc (chromium) pinned because its build seems more finicky.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
https://github.com/keroserene/go-webrtc/issues/23
https://webrtc.org/native-code/development/
https://webrtc.org/native-code/development/prerequisite-sw/
|
| | |
|
| |
|
|
|
| |
7z does not currently extract file permissions from the dmg files
so we also extract the old mar file to copy the permissions.
|
| |
|
|
|
|
| |
The subkey that signed PyCrypto back in the days expired. We fall back
to the SHA256 check (which we already did in addition to the signature
check).
|
| |
|
|
|
|
|
|
|
|
| |
We ecountered on some machines failures to reproduce OS X builds.
The investigation seems to indicate that our old toolchain plays a
crucial role in this.
This patch replaces this old toolchain with a more modern clang/cctools
for building the tor part of the bundle. Resorting to `faketime` again
was necessary to cope with inserted timestamps.
|
| |
|
|
|
|
| |
This reverts commit 0348263efd1cb8a9eca8737f3dc7734ef75e0966.
Firefox is not ready for GCC 6 yet. See #20381 for further details.
|
| | |
|
| |
|
|
|
|
|
|
| |
When updating the OSX incremental MAR files after running dmg2mar, we check
that both the old and the new mar files contain code signing.
We don't check this while creating the incremental MAR files during the
build.
|
| |
|
|
|
|
|
|
|
|
| |
Bumping GCC to 6.2.0 is fixing our EMET incompatibilities. The exact
change on GCC's side causing this remains unclear, though.
We need to backout GCC changes that are causing a mingw-w64 build to
fail. For details see the bug report in GCC's bugtracker:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77459
|
| |
|
|
|
|
|
|
|
| |
LeifEricson (blocked): Port 50000 --> 50001
Greenbelt (unblocked): Port 5881 --> 7013
Mosaddegh (blocked): Port 2934 --> 9332
MaBishomarim (blocked): Port 2413 --> 7920
JonbesheSabz (blocked): Port 1894 --> 4148
Azadi (unblocked): Port 4319 --> 6041
|
| |
|
|
|
|
| |
This version has a fix for https://bugs.torproject.org/20030.
(meek-client-torbrowser failing to kill is subprocesses, which was a
problem whenever meek-client-torbrowser was built with Go 1.6 or later.)
|
| | |
|
| |
|
|
|
|
|
|
|
|
| |
Don't touch the incremental mar files for the other platforms when
updating the OSX incremental mar files after adding code signing.
We add a MAR_SKIP_EXISTING environment variable which indicates to the
gen_incrementals script that existing incremental mar files should not
be regenerated. The old incremental mar files for OSX are removed by
the dmg2mar script.
|
| | |
|
| | |
|