Only consume an AVPacket when all the samples have been read.
When the rate of samples output is limited (by the default value
of max_samples), consuming the first packet immediately will cause
timing problems:
- The first packet with PTS 0 will output 4608 samples and be
consumed entirely
- The second packet with PTS 64 will output the remaining samples
(typically, a lot, that's why max_samples exist) until the decoded
samples of the first packet have been exhausted, at which point the
samples of the second packet will be decoded and output when
av_decode_frame is called with the next packet).
That means there's a PTS jump since the first packet is 'decoded'
immediately, which can be seen with avplay or mplayer: the timing
jumps immediately to 6.2s (which is the size of a packet).
Sample: http://streams.videolan.org/issues/6348/Goldwave-MAClib.ape
Signed-off-by: Justin Ruggles <justin.ruggles@gmail.com>
The i686 feature really is a CPU feature and should be handled as such.
The cpunop dependency on i686 should be expressed with a standard _deps
declaration instead of a manual test.
When streaming to limelight, the app name is either a full
"appname/subaccount" or "appname/_definst_". In the latter case,
the app name can be simplified into simply "appname", but the
authentication hashing assumes the /_definst_ still to be present.
Signed-off-by: Martin Storsjö <martin@martin.st>
When using AV_NOPTS_VALUE (which expands to INT64_C(0x8000000000000000))
as union initializer for a double field, the c99 converter needs to
interpret this constant when filling the union initializer, and it is
interpreted as a positive value.
When converting AV_NOPTS_VALUE to a double, MSVC 2010 ends up with
the same positive value as the c99 converter, while MSVC 2012 gets
a negative value.
This results in an infite loop in various FATE tests on MSVC 2012.
Signed-off-by: Martin Storsjö <martin@martin.st>
This reverts commit 4622f11f9c.
The compiler should be able to do the dead code elimination now
without this when the cond_* names point directly to the real
functions instead of to local function pointers.
Signed-off-by: Martin Storsjö <martin@martin.st>
This reduces the call overhead slightly. More noticeably, it
restores the earlier (unintended?) feature that condition variable
functions work just fine even if w32thread_init() hasn't been called.
This was broken as a side effect of 4622f11f9, if explicitly targeting
Vista+.
This makes w32threading work in VP8 again, if targeting Vista+.
Signed-off-by: Martin Storsjö <martin@martin.st>
If a client tries to read the file while it's being updated, the client
would get an incomplete manifest. Instead write to a separate temp file
and atomically rename it to replace the previous one.
Signed-off-by: Martin Storsjö <martin@martin.st>
The element was only being written when the value == 1. But the default
value of this element is 1, so this has no useful effect. This element
needs to be written when the value == 0.
Signed-off-by: Anton Khirnov <anton@khirnov.net>
On failures in the write_trailer function, we could also ignore
the errors and try to finish the file despite these errors (which
would only leave an incomplete chapters track). It's probably better
to signal the error clearly to the caller though (and if this
function failed there's no guarantee that there's enough memory to
finish the trailer either).
Signed-off-by: Martin Storsjö <martin@martin.st>
This both allows factoring out size check for both MetaSound and TwinVQ-VQF
decoders and fixes the situation when there are several MetaSound frames
stuffed together (that happens in 8kHz @ 8kbps MetaSound in ASF for example).
