This fixes a regression in 4ed5ac50.
Signed-off-by: Martin Storsjö <martin@martin.st>

The vec_splat() intrinsic requires a constant argument for the
element number, and the code relies on the compiler unrolling
the loop to provide this.  Manually unrolling the loop avoids
this reliance and works with all compilers.
Signed-off-by: Mans Rullgard <mans@mansr.com>

Signed-off-by: Mans Rullgard <mans@mansr.com>

Enable dead store elimination.  The last few releases work no worse
with this flag than without.  Older versions failed to build some
source files when using this flag.
Signed-off-by: Mans Rullgard <mans@mansr.com>

Tested to work with the gstreamer depacketizer.
Signed-off-by: Martin Storsjö <martin@martin.st>

Not all details are implemented, but it's enough for proper
playback as long as there is no packet loss.

Tested to work with the packetizer in gstreamer (which although
uses a different codec name, to clarify that it is still a spec
draft).
Signed-off-by: Martin Storsjö <martin@martin.st>

Signed-off-by: Mans Rullgard <mans@mansr.com>

Rename the called dsp init functions to *_init_x86.

The tag @table expects @item to mark entries.

avconv provides an option to set it externally and there isn't a way for
the outer demuxer to report the inner demuxer flag.

Solves a bad interaction between mpegts and x264, but requires
additional setting for the user.

When there is only 1 channel, the planar and interleaved formats of the same
data type should be treated as identical.

When there is only 1 channel, the planar and interleaved formats of the same
data type should be treated as identical.

Should fix fate-acodec-adpcm-ima_wav with several compilers.

Signed-off-by: Martin Storsjö <martin@martin.st>

It is used for multiple codecs.

An alpha specifier outside the valid range results in a conversion from
double to long with undefined result.  Range-checking the double and
only converting it after it passes avoids this.

Fixes fate-parseutils errors on some systems.
Signed-off-by: Mans Rullgard <mans@mansr.com>

Signed-off-by: Diego Biurrun <diego@biurrun.de>

This helps tracking down the cause if this happens to be an issue.
Signed-off-by: Martin Storsjö <martin@martin.st>

We don't need to return an error if the directory already existed.
We make sure that the directory actually is properly writeable later
within ism_write_header by writing a manifest anyway.
Signed-off-by: Martin Storsjö <martin@martin.st>

This allows failing cleaner and earlier if unable to write to
the output directory.
Signed-off-by: Martin Storsjö <martin@martin.st>

This allows calling write_manifest from ism_write_header without
a forward declaration.
Signed-off-by: Martin Storsjö <martin@martin.st>