It's almost as if OGG Vorbis is a terrible terrible terrible file format.
kepstin
vorbis is fine, it's just the ogg ;)
I guess they just didn't bother to put support for padding into the format?
huh, it looks like there are some patches to some tools to use padding in the comments.
there's an annoying thing with the flac tag writer in picard - if it thinks there's /too much/ padding for the comments in a flac file, it'll rewrite the flac file to reduce the padding.
and the amount of padding that the flac encoder includes by default is above the limit that picard has :/
so picard completely rewrites every freshly encoded flac I tag with it :(
nikki
enter a ticket then :P
kepstin should probably file a bug for that, but it's more likely a mutagen issue than in picard directly?
bitmap would probably know
or luks, but he's probably asleep
jacobbrett joined the channel
jacobbrett joined the channel
jacobbrett joined the channel
ianmcorvidae
kepstin-laptop: perhaps http://mutagen.sourcearchive.com/documentation/... or so is to blame? It's not supposed to change the size of PADDING blocks but perhaps the way picard's writing tags results in multiple blocks that then must be merged?
kepstin-laptop
ianmcorvidae: I dunno, but the flac encoder writes 8kb of padding by default, and mutagen always reduces that to 1kb or less padding (after tag writing)
I thought that _load maybe wasn't reading in padding but I was wrong
meh, probably better to just ask luks/bitmap later, I don't know this codebase :)
kepstin-laptop
heh
kepstin-laptop has been looking at it, but nothing's jumping out
it looks like there shouldn't ever be a case where it makes the header section smaller, there simply isn't code to do that
it has code to handle the existing space in the header being too small, that's what the call to insert_bytes on line 674 is for.
so the code path should be, given a simple example of "old vorbiscomment=1kb old padding=7kb", doing an update that causes new vorbiscomment=2kb, it should add 1kb padding to the end, giving 2kb comment + 8kb padding, notice that that's over the available space, and cut the padding down to 6kb.
ianmcorvidae
yeah :/\
possibly would be worth going through this at the python command-line with mutagen
kepstin-laptop
yeah.
there is literally no place in flac.py that could cause the flac file to get smaller
ianmcorvidae
at least isolate the problem to mutagen vs. picard, since we've been assuming mutagen but it looks like mutagen is clean
kepstin-laptop
the insert_bytes() function will give an assertion failure if passed a negative number, and there's no call to delete_bytes
Could an auto-editor expediate this process somehow? Either by first accepting the merge before the track list change, or by accepting the track list change before the merge?
nikki
auto-editors can only accept things they can auto-edit themselves, so not merges
the latter should work though
cosares
I'll give it a go, and post the edit here
nikki
they should probably be fixed first anyway, since otherwise merging will merge the wrong recordings
cosares
Do I have to rename the tracks (and swap recordings) or is there a re-order tracks option?
reosarevok
cosares, in the advanced traclist, just change the track numbers