My use case is to update an existing CUE file, not create one. What’s yours?
I was trying to create new cuesheets that were compatible with CUETools. It’s been a few years, but I think I ran into the same problem you describe of CUETools refusing to consume the generated cuesheets. One potential problem is that the plugin writes fully qualified pathnames. I updated those before importing into CUETools, but iirc, it still didn’t work.
I wanted to convert DJ mixes to/from monolithic files. DJs sometimes release a mix as a single file, and publish the track times. After adding the tracked mix to MusicBrainz (as a pseudo-release), it becomes possible to create a cuesheet to split a monolithic mix into tracks. Conversely, if I ever had to use a media player that didn’t support gapless playback, I could re-encode a continuous mix as a single file.
Below are a few notes about the issues you mentioned:
the values were not 00:00:00, but rather the cumulative length of the tracks to that point. Slightly off versus the actual index values from the cue file
Isn’'t that normal (and necessary) for cuesheets? But yes, the offsets in a generated file can differ from those in your original cuesheets. It’s not necessarily the result of a code error, but this is where more analysis of the plugin would clarify matters.
on a folder with a single large .wav file (the whole CD) it forgot to include the file name!
I don’t think the plugin supports monolithic files. In any case, if you invoke it from a release with no files attached to it in Picard, it omits the FILE line completely, since it doesn’t know what file(s) the cuesheet is meant for.
it overwrites the existing .cue file which means some information such INDEX and ISRC is lost
Since the plugin generates a new cuesheet, it might help to use a diff tool to merge any ISRCs, comments, pre-gap indices, etc. into the generated file. Meld (Linux, Windows) and WinMerge (Windows) are both free and open-source.