Byte order in ogg vorbis headers
Ryan Rix <[email protected]>
| Newsgroups | gmane.comp.multimedia.ogg.vorbis.general |
|---|---|
| Message-ID | <[email protected]> |
Hello all, I am currently writing an application for PalmOS handheld devices that organizes media files on a memory card and allows the user to easily create playlists. The tracks I am using for testing (totalling about 1.5 GB) are nearly all in Ogg Vorbis format. Throughout my testing I have had very little trouble in parsing the comment fields from mp3, asf and ogg files. That is, all except for one or two files. PalmOS runs on m68k hardware (or emulated m68k on newer devices) so the byte order is big endian. The issue I seem to be having is that when parsing the comments I use a for loop limited by the number of fields, which is a value stored within the ogg file, one or two bytes before the comments begin (I don't have my code handy to check atm) According to the Vorbis docs this is stored as unsigned integer of 32 bits. I assume that it is x86 little endian and thus byte swap it using an OS provided macro. On these one or two files I've encountered the application seems to hang. However on debugging the application I discover that the number of fields is exorbinately large. After byte swapping the value by hand I discover that it represents the correct number of fields for that file. My question is, I guess, is there a standard endianness for Ogg encapsulated Vorbis files' meta data? Thanks and best regards, Ryan Rix TamsPalm - The PalmOS Blog My heart is human, my blood is boiling, my brain: IBM