Monkey's Audio Codec (.ape) playback not working

[rathann]

Description of problem:

Playback of .ape files doesn't work.

Version-Release number of selected component (if applicable):

2.0-4.fc44

How reproducible:

Always

Steps to Reproduce:

1. Run aqualung
2. Open any .ape file. (sample: https://samples.mplayerhq.hu/A-codecs/lossless/luckynight.ape)
3. Play

Actual results:

[ape @ 0x7fbe14603dc0] output buffer size is too small for the current frame (9216 < 18432)

Expected results:

Working playback.

Additional info:

This happens with both mac-10.18 and the latest mac-12.50. Also reproducible on Fedora 43 with aqualung-1.2.

Downstream bug report: https://bugzilla.redhat.com/show_bug.cgi?id=2446245 .

[tomscii]

You have an Aqualung binary built without the native Monkey's Audio codec. The warning you get is from the lavc decoder that tries to open everything not claimed by other decoders.

I just tested Aqualung with the native MAC library built from source (download the source SDK from https://www.monkeysaudio.com/developers.html, build it with cmake according to the instructions in the Readme.txt and do a make install, now Aqualung's configure script will detect the presence of the codec). It works perfectly.

[rathann]

You have an Aqualung binary built without the native Monkey's Audio codec.

I beg to differ:

https://kojipkgs.fedoraproject.org//packages/aqualung/2.0/6.fc44/data/logs/x86_64/build.log

configure: Libav / FFmpeg Support
checking for libavcodec libavformat libavutil... yes
configure: Monkey's Audio Codec Support
checking for CreateIAPEDecompress in -lMAC... yes

The warning you get is from the lavc decoder that tries to open everything not claimed by other decoders.

Ok. I haven't looked at the code, but it looks like it has higher priority than the native libMAC decoder, then.

I just tested Aqualung with the native MAC library built from source (download the source SDK from https://www.monkeysaudio.com/developers.html, build it with cmake according to the instructions in the Readme.txt and do a make install, now Aqualung's configure script will detect the presence of the codec). It works perfectly.

Can you try with both MAC and FFmpeg support compiled in?

[tomscii]

Interesting. Can you confirm seeing Monkey's Audio Codec included in the About box of the binary you are actually running?

I do have both MAC and lavc compiled in, and the MAC decoder has priority (as it should - lavc is the last decoder to try as a last resort [1]). Which is why I'm puzzled about your report.

[1] Order of decoders:
https://github.com/jeremyevans/aqualung/blob/master/src/decoder/file_decoder.c#L74

[tomscii]

From the build log you linked to:

configure: WARNING: unrecognized options: --with-mac

That should be --with-MAC

Further down:

configure: Monkey's Audio Codec Support
checking for CreateIAPEDecompress in -lMAC... yes
./configure: line 12303: ,: command not found

That does not look right to me, even though the configure listing includes MAC as "yes (found)".

[rathann]

Interesting. Can you confirm seeing Monkey's Audio Codec included in the About box of the binary you are actually running?

Yes. There's both Monkey's Audio Codec and LAVC.

I do have both MAC and lavc compiled in, and the MAC decoder has priority (as it should - lavc is the last decoder to try as a last resort [1]). Which is why I'm puzzled about your report.

[1] Order of decoders: https://github.com/jeremyevans/aqualung/blob/master/src/decoder/file_decoder.c#L74

OK...

From the build log you linked to:

configure: WARNING: unrecognized options: --with-mac

That should be --with-MAC

Ok. However, correcting that doesn't change...

Further down:

configure: Monkey's Audio Codec Support
checking for CreateIAPEDecompress in -lMAC... yes
./configure: line 12303: ,: command not found

...the output above.

That does not look right to me, even though the configure listing includes MAC as "yes (found)".

Agreed. However, the m4 macros in configure.ac make my head hurt:

AQUALUNG_DETECT([MAC], [Monkey's Audio Codec],
    [AC_CHECK_LIB([MAC], [CreateIAPEDecompress],
        [
          AQUALUNG_FOUND([MAC], [-lMAC -lstdc++]),
          AS_CASE([$host_os],
            [cygwin*|mingw*], [AC_DEFINE([PLATFORM_WINDOWS], [1], [Defined if Windows])],
            [darwin*], [AC_DEFINE([PLATFORM_DARWIN], [1], [Defined if Darwin])],
            [linux*|*bsd*], [AC_DEFINE([PLATFORM_LINUX], [1], [Defined if Linux/BSD])])
        ],
        [AQUALUNG_MISSING([MAC],
            [Monkey's Audio Codec support requires libMAC])],
        [-lstdc++])])

There's probably an extra comma there somewhere, because the resulting configure script has this:

    if test "x$with_MAC" = xdetect
then :
  with_MAC='yes (found)'
fi
    MAC_LIBS='-lMAC -lstdc++'

,
          case $host_os in #(
  cygwin*|mingw*) :

[tomscii]

Regarding the configure.ac quirk, there is a fix committed about a year ago:
ecb4cd1

Can you try compiling the actual upstream source code (latest master) locally and see if you can reproduce the issue?