I think the problem was that I only introduced delay in one direction, and the estimation assumes that the RTT is evenly divided between the send direction and the receive direction, which was true for the old test.
BUG=chromium:576246
R=pbos@webrtc.org
Review URL: https://codereview.webrtc.org/1577853005 .
Cr-Commit-Position: refs/heads/master@{#11233}
This let's us use them to configure them when using WebRTC as an external library. One use case where this is necessary is in the Android OS.
Review URL: https://codereview.webrtc.org/1538643004
Cr-Commit-Position: refs/heads/master@{#11231}
It looks to me like targets :rtc_base_approved is logically a subset of
:rtc_base, and so any targets depending on :rtc_base expect to also get
access to the headers in :rtc_base_approved.
Thus I think it's appropriate for :rtc_base to have :rtc_base_approved in
public_deps, so that `gn check` will permit this without clients having to
explicitly depend on both.
NOTRY=True
Review URL: https://codereview.webrtc.org/1578833002
Cr-Commit-Position: refs/heads/master@{#11227}
There's no reason not to use std::move instead now that we can use the
C++11 standard library.
BUG=webrtc:5373
Review URL: https://codereview.webrtc.org/1531013003
Cr-Commit-Position: refs/heads/master@{#11225}
The unit test will run the pure C denoiser and SSE2/NEON denoiser (based
on the CPU detection) and compare the denoised frames to ensure the bit
exact.
TBR=tommi@webrtc.org
BUG=webrtc:5255
Review URL: https://codereview.webrtc.org/1492053003
Cr-Commit-Position: refs/heads/master@{#11216}
The general constraints on number of channels for AudioProcessing is:
num_in_channels == num_out_channels || num_out_channels == 1
When Beamforming is enabled and additional constraint was added forcing:
num_out_channels == 1
This artificial constraint was removed by adding upmixing support in CopyTo, since it was already supported for the AudioFrame interface using InterleaveTo.
Review URL: https://codereview.webrtc.org/1571013002
Cr-Commit-Position: refs/heads/master@{#11215}
Defining use_third_party_h264 directly, and indirectly defining use_openh264 (from third_party/openh264) and ffmpeg_branding (from third_party/ffmpeg).
These will be used in a follow-up CL that adds an encoder and decoder implementation.
The flags are added in this CL so that they can be used by trybots/waterfall bots in GN without "Build argument had no effect" errors. Equivalent GYP changes are also added.
BUG=468365
Review URL: https://codereview.webrtc.org/1575913003
Cr-Commit-Position: refs/heads/master@{#11204}
This meant splitting "transport_options" into audio/video/data options,
for when creating the answer, and giving "GetSslRole" a "transport_name"
parameter so we can retrieve the current role on a per-transport basis.
BUG=webrtc:4525
R=pthatcher@webrtc.org
Review URL: https://codereview.webrtc.org/1516993002 .
Cr-Commit-Position: refs/heads/master@{#11192}
* Better param names
* Avoid using negative values for (bogus) placeholder channel counts (mostly in tests). Since channels will be changing to size_t, negative values will be illegal; it's sufficient to use 0 in these cases.
* Use arraysize()
* Use size_t for counting frames, samples, blocks, buffers, and bytes -- most of these are already size_t in most places, this just fixes some stragglers
* reinterpret_cast<int64_t>(void*) is not necessarily safe; use uintptr_t instead
* Remove unnecessary code, e.g. dead code, needlessly long/repetitive code, or function overrides that exactly match the base definition
* Fix indenting
* Use uint32_t for timestamps (matching how it's already a uint32_t in most places)
* Spelling
* RTC_CHECK_EQ(expected, actual)
* Rewrap
* Use .empty()
* Be more pedantic about matching int/int32_t/
* Remove pointless consts on input parameters to functions
* Add missing sanity checks
All this was found in the course of constructing https://codereview.webrtc.org/1316523002/ , and is being landed separately first.
BUG=none
TEST=none
Review URL: https://codereview.webrtc.org/1534193008
Cr-Commit-Position: refs/heads/master@{#11191}
This allows the test to create its own transports if it, for instance, needs to do demuxing.
BUG=webrtc:5416
Review URL: https://codereview.webrtc.org/1573453002
Cr-Commit-Position: refs/heads/master@{#11187}
To ease use of WebRTC in other codebases, update some macros
to match glibc's ansidecl.h, which uses double-underscores for attributes.
NOTRY=True
Review URL: https://codereview.webrtc.org/1571653002
Cr-Commit-Position: refs/heads/master@{#11185}
Ask the OS for the mic volume every 1 second rather than with every 10
ms chunk. The previous behavior was consuming ~2% of the CPU load of
a voice engine call, and is now negligible.
This is consistent with the webrtc Windows Core Audio implementation,
as well as the Chromium Mac implementation:
https://code.google.com/p/chromium/codesearch#chromium/src/media/audio/agc_audio_stream.h
TEST=voe_cmd_test with AGC continues to work well on Mac.
Review URL: https://codereview.webrtc.org/1564223002
Cr-Commit-Position: refs/heads/master@{#11182}
Tests were failing on android with new libvpx.
vp9 speed setting was changed to 8 recently and some recent changes
in libvpx require update for the tests to pass.
TBR=stefan@webrtc.org
BUG=webrtc:5401
Review URL: https://codereview.webrtc.org/1569903002 .
Cr-Commit-Position: refs/heads/master@{#11173}
Add audio send and receive streams to CallTest and call the necessary voice engine APIs for the streams to be usable. Verifies the implementation by adding a simple test which monitors outgoing packets and checks that both audio and video is being sent with transport sequence numbers.
Audio streams are using a fake audio device with file input.
The CallTest implementation is to a big degree based on call_perf_tests.cc and should in the future replace a lot of that code.
R=pbos@webrtc.orgTBR=kjellander@webrtc.org
BUG=webrtc:5263
Review URL: https://codereview.webrtc.org/1542653002 .
Cr-Commit-Position: refs/heads/master@{#11171}
Removes multiple index lookups to generated_fec_packets_ speeding up
FecTest.FecTest with >2x in both Debug and Release, improving
performance but also readability.
On Debug this means that the slowest test in modules_tests now takes
~15-20 seconds instead of 50+ seconds, reducing the overall bottleneck.
BUG=webrtc:4712
R=stefan@webrtc.org
Review URL: https://codereview.webrtc.org/1552563003 .
Cr-Commit-Position: refs/heads/master@{#11166}
Update test implementation (test/histograms.h) to be more similar a real implementation (where histogram get functions return a Histogram pointer). Add check that the name of a histogram does not change.
BUG=webrtc:5283
Review URL: https://codereview.webrtc.org/1528403003
Cr-Commit-Position: refs/heads/master@{#11161}
