#99 buffer truncates songs at Mac OS X

mpg123 (104)

When using the option -b for buffering song then the last second (round about that) is truncated in playback with mpg123-1.8.0.

Darwin xxx 9.7.0 Darwin Kernel Version 9.7.0: Tue Mar 31 22:54:29 PDT 2009; root:xnu-1228.12.14~1/RELEASE_PPC Power Macintosh

(Tested with test suite from http://web.student.tuwien.ac.at/~e0325716/mpg123-test.tar.gz\)


  • Andreas Neustifter

    BTW: Linux is fine. I do not test windows.

  • Thomas Orgis

    Thomas Orgis - 2009-06-13

    Interesting... is mpg123 1.7.3 fine? I fixed an isse of buffer swallowing audio at beginning... now it's only fair to have it missing at the end...
    But really, it would be good to know if it is a regression from earlier mpg123 versions or not.

  • Andreas Neustifter

    This also happens with 1.7.3. (With 1.7.3 the truncating at start also happens, but as you fixed it not anymore with 1.8.0.)

  • Thomas Orgis

    Thomas Orgis - 2009-07-28

    Hm, please confirm: This issue is still persisting with current trunk (to-be 1.9.0)?

    Perhaps we need to fix some buffering stuff on the coreaudio output, like with win32.

  • Andreas Neustifter

    First of all: I was trying to compile the trunk under MacOS (I used only the tarballs before) and autogen fails with not finding libtool. I will open a new bug and attach a patch.

    Secondly the build fails in synth_altivec.c with the following messages:
    > In function ‘synth_1to1_altivec’:
    > synth_altivec.c:299: error: ‘vector’ undeclared (first use in this function)
    > synth_altivec.c:299: error: (Each undeclared identifier is reported only once
    > synth_altivec.c:299: error: for each function it appears in.)
    > synth_altivec.c:299: error: syntax error before ‘float’
    > synth_altivec.c:305: error: ‘vclip’ undeclared (first use in this function)
    > synth_altivec.c:306: error: ‘vzero’ undeclared (first use in this function)
    > synth_altivec.c:307: error: ‘vshift’ undeclared (first use in this function)
    > synth_altivec.c:309: error: ‘vmax’ undeclared (first use in this function)
    > synth_altivec.c:309: error: syntax error before ‘float’
    > [...]

    Any ideas?

  • Andreas Neustifter

    Okay, I patched the code and its works, I'm not sure tough, anyways, I opened a new bug for the build problems.

    I'm afraid the trucating of the end of the audio output persists with the current trunk. What can I do to help? Would you like an audio recording of the problem? Do you have access to a PPC Mac?

    Some further system information:

    uname -a:
    "Darwin xxx 9.7.0 Darwin Kernel Version 9.7.0: Tue Mar 31 22:54:29 PDT 2009; root:xnu 1228.12.14~1/RELEASE_PPC Power Macintosh"

    System Profiler:
    Model Name: iBook G4
    Model Identifier: PowerBook6,7
    Processor Name: PowerPC G4 (1.5)
    Processor Speed: 1.33 GHz
    Number Of CPUs: 1
    L2 Cache (per CPU): 512 KB
    Memory: 1,5 GB
    Bus Speed: 133 MHz
    Boot ROM Version: 4.9.3f0
    Serial Number (system): 4H5520EMSE4
    Sudden Motion Sensor:
    State: Enabled
    Version: 1.0

  • Thomas Orgis

    Thomas Orgis - 2009-07-28

    We are talking of the live playback or also of writing to a WAV file? Not since too long ago, you can write to plain files with the buffer (even if it does not make much sense, it is useful for debugging).
    Can you try if you get the full output when doing mpg123 -w test.wav -b 4096 test.mp3 ?
    Of course this could be a Heisenbug that manifests only with real output.

    I need to get an angle on how this can be spefific to MacOS ... can you also try other output modules than CoreAudio? Like, SDL? Portaudo?

  • Andreas Neustifter

    Yes, its live playback with
    >$ mpg123 -C -b 100 test.mp3

    The problem does not occur with wav output. So no, with
    >$ mpg123 -w test.wav -C -b 100 test.mp3
    the wav file is complete.

    I do not have any other audio output modules available right now, i tried openal but this gives me only crackling in the speakers and no sound.

    (The other system configuration can be found with the attached files from bug 2828682.)

  • Thomas Orgis

    Thomas Orgis - 2009-07-29

    Hm, openal not working is a bug ... please report this (and, best, drop a mail to the list so that Taihei sees it... he added the OpenAL output). It would really help to pin it to the Coreaudio output... so, is it possible to install another audio library (like SDL)?

  • Andreas Neustifter

    Sorry for lagging, RL also takes its time...

    I have tried it with the libsdl from MacPorts (the Runtime Libraries from libsdl.org did not work) and SDL is not working properly even without buffering:

    $> mpg123 -o sdl test.mp3

    truncates the last second or so.

    I will report this and the OpenAL bugs immideatelly.

  • Thomas Orgis

    Thomas Orgis - 2013-04-21

    Been a long time. Could you re-check this with current mpg123? The coreaudio driver has been touched, but not specifically for that issue. But I fixed the truncation with SDL (at least on Linux) in the latest snapshot: http://mpg123.org/snapshot .

    If it's still truncating on MacOS, I'd like to get this fixed before relasing 1.16.0 .

  • Andreas Neustifter

    Sorry, I'm not using mpg123 on Mac OS anymore. I guess you could just close the bug and reopen it if needed.

  • Thomas Orgis

    Thomas Orgis - 2013-10-16

    Hm, what's the state? SDL is fixed ... the buffer thing ... ?

  • Thomas Orgis

    Thomas Orgis - 2016-09-24
    • status: open --> closed
    • Group: --> 0.68

Log in to post a comment.