Add bigendian support to mythmusic

[roltman@…]

The various decoders used in mythmusic were designed to use only little endian processors. The powerpc is a big endian processor and all music would create static or noise when played.

Attatched is a "svn diff" of the changes I have used to get it working for: ogg and mp3 (tested) and I also patched the flacdecoder.cpp but did not test it.

I did not check any of the encoding features to see if there is an endian problem there.

[roltman@…]

svn diff of my patch for big endian playback

[Isaac Richards]

This really needs to be autodetected, not an option to configure.

[roltman]

I'll see if I can dig up a configure option. Most of the research I did on this was based off mpg321 & it does autodetect this so I'll submit a new patch for configure.

[roltman]

Added <endian.h> to files and removed configure option

[magnus@…]

Why is this bug so low on priority/severity? MythMusic is actually completely unusable on big endian platforms without this fix, since all you get to here is white noise!

I can confirm that the patch solves the problem and seems to have no other problems. It seems to me to be of quite low risk, and of extreme importance for big endian platforms. I'm a bit surprised it didn't go into 0.19.

[bhulsken@…]

fix mythmusic playback for bigendian systems

[anonymous]

I've adapted roltmans patch to apply agains the 0.19-fixes (revision 9093)

I hope I'm more or less following the right procedure for submitting patches..

Anyway, it works for me on a mac-mini ppc, instead of noise, I can hear music. Can it be applied to the 0.19-fixes branch? It seems trivial, and at the moment mythmusic for bigendian systems is clearly broken. I don't really understand why this ticket is rated as trivial/low, but perhaps because nobody uses big-endian systems.

[bhulsken@…]

After some more days of testing, I can testify that this patch works perfectly. Unfortunately I can't test for AAC, since faad2 wont compile on bigendian.

I've changed the severity of this ticket to high. Since this patch fixes a bug in mythmusic that makes mythmusic completely unusable on ppc, classifying it as low severity is a bit of an understatement.

I don't want to sound to impatient, but is there a reason this doesn't get committed? The patch is almost a month old, and this bug prevents all ppc users from using MythMusic. If it's too ugly or there's something else wrong with it, give me a clue, and perhaps I can rewrite it a bit.

[Nigel]

Finally had time to install Linux on a PowerPC machine (Ubuntu on a PowerBook? G4). Am now up to compiling plugins. Assuming all goes well, this will be in soon.

[Nigel]

The existing MythTV code uses a test in configure to set WORDS_BIGENDIAN (or not). The supplied patches use endian.h and a #if BYTE_ORDER == BLAH_ENDIAN.

There is nothing intrinsically wrong with the patches, but a lot of machines don't have endian.h (e.g. Mac OS X has machine/endian.h instead), so I will use mythtv/mythconfig.h and the slightly simpler #if instead.

[Nigel]

(In [9718]) Endianness fixes for some MythMusic decoders. Fixes #1024.

[Nigel]

(In [9719]) Big Engian fixes for MythMusic - in fixes branch too. See #1024

[Nigel]

(In [9720]) MythMusic Big Endian fix. Use a #ifdef instead of an #if. See #1024