This caused us to build everything 2-4 times, now we only build it 1-2
times (depending on selection of static/shared builds). The disadvantage
is that now it's more up to the reviewer to ensure modularity (we still
catch it in the bazel build, though).
This change does not include the addition of VP9. We do that in a
separate pull request.
Changes:
* fix the video bug (video frames larger than 65KBytes) by sending full
frame length in alternate header field
* improve video frame reconstruction logic with slots
* configure video encoder and decoder to be multihtreaded
* set error resilience flags on video codec
* change encoder and decoder softdeadline
The logic and behaviour remain the same, but PEER_EXIT and PEER_JOIN are
merged into LIST_CHANGED. This allows clients to take a first step to
adapt to the API change without us having to actually submit the full
change, yet.
Also, add an auto-test for bootstrap and for LAN discovery.
Bootstrap is never tested otherwise, and LAN discovery is a prerequisite
for everything else. Having these two tests lets us rule out or identify
LAN discovery as a possible cause for test failures.
Instead of hard-erroring when we don't find dependencies, we just don't
build the things requested and print a warning. This is less annoying to
users because we enable a bunch of things by default. This way, we can at
least build something with the default "cmake" invocation.
Also reduce number of people in conference to 5, because on Circle CI the
test times out trying to connect more than 6 or 7 people. The persistent
conferences PR will improve this so we can set it much higher then.
Since trusty has more up-to-date packages, we can remove some of the
custom install code. Also, we're not using hstox at the moment, so there
is no need to slow down the builds for now.
This also adds RTPCapabilities and a header field to tell the receiver
about capabilities used in encoding this frame. It is intended to contain
settings relevant to the current frame being sent.
follow TokTok#731. This commit
completely removed all things in namespace bit_rate, and deprecated
functions are to be added back in another commit. set_xxx() is treadted
as a property of namespace audio&video, same as bit_rate change event.
toxav_basic_test is fixed, either.
Also got rid of two VLAs. They are overused a bit in toxcore. In
irc_syncbot, the array was uninitialised and then filled by a recv system
call. This can cause uninitialised reads if recv doesn't fill the entire
array. It could not cause out of bounds read directly, because a
NUL-terminator was in place, but both cases are undefined behaviour.
Fixes#572.
As discussed in the issue, there's a risk that toxcore may not hold the
maximum bitrates libvpx supports, if toxcore insists on using integer
type. I initially proposed to have another flag in set(), so that we can
use unsigned type instead. iphydf came up with a better solution, that is
splitting the original functions, one for audio, one for video. Now, we
could safely replace int32_t with uint32_t.
Also: clean video_bit_rate_invalid()
Though this is not a part of issue #572, as it's used in the
toxav_bit_rate_set(), i cleaned the code. As mannol said, there should be
a check. Uint32_t is large enough to hold the maximum bitrates libvpx
supports, but user may pass a value larger than uint while smaller than
uint32_t. Thanks to the reminding from nurupo, it's no longer a stub
function.
Bitrate error enums are shared for both audio and video
https://github.com/TokTok/c-toxcore/pull/578#issuecomment-360095609, just
as iphydf said.
