ref: 2e85824bfac2a2ff26e62acaa898328991e7521c
dir: /BUGS/
• unsure if seeking on a file works correctly • current treatment of lossy sources is less than ideal • everything will get resampled to 44100Hz/16bit/2channel • the codebase is a mess • reading concurrently from the same fid might break decoding logic, or be unecessarily slow (does that ever happen?) • the parser might be leaking memory (it strdup()s the strings before passing to yacc, but we strdup() some of them again when setting the sheet fields) • zuke with wav keeps requesting the file header at offsets 0, 4 and 8; never starts playing • some parser/cuefs behavior might be non-standard or undesirable - for example, there's currently no way to split tracks without including the pregap either at the end of the last track or beggining of current one • splitting tracks from → to the same format could some times be done more efficiently, skipping the decode (and resample) → encode process; this could be especially beneficial to lossy formats since there would be no reencoding. • some cuesheets have metadata (that would otherwise have nowhere to be in the sheet) stored in comments, in a somewhat semistandard manner, i.e. REM GENRE "Progressive Rock" REM DATE 2002 it probably wouldn't hurt to try parsing some of those; not a high priority at the moment • cuefs does not respect PREGAP and POSTGAP tags on cuesheets • zuke's mkplist dislikes cuefs, somehow (trying to mkplist on a cuefs mountpoint doesn't even generate 9p rpcs, something's really wrong) • we likely don't handle "syntatically valid but semantically invalid" cuesheets very well