removing or archiving old data, rolling data files

[meonkeys]

I'm curious about the Taskwarrior data files backlog.data, completed.data, pending.data, and undo.data. Are these "permanent" (must be left as-is), or can these be "rolled" like logfiles? I rolled the undo.data file by renaming and compressing the current one and a new undo.data file was automatically created. I forget where I got the idea that it was OK to do this.

Is there some way to easily save space by rolling, archiving, or even deleting old data? Sorry if I missed this in the docs or a mailing list/github discussion.

[djmitche]

They're all a little different, despite having the same extension:

• pending.data contains only pending tasks, and, assuming you keep your pending tasks at a relatively constant count, stays a reasonable size.
• completed.data contains all non-pending tasks, and grows without bound. It can be deleted, if you're OK with losing all that history. Deleting may interact in "interesting" ways with sync.
• backlog.data contains changes not yet transmitted to taskd by task sync. If you sync regularly, it should always be quite small.
• undo.data contains all changes made to all tasks. It is used to support task undo but also to generate the history shown in task info and a few other places. It grows without bound. It can be safely deleted, if you're OK not having that history in task info.

None can really be "rolled" like a logfile -- they're not log files. I don't see much point to saving the file contents elsewhere if you're going to delete the files Taskwarrior uses.

Taskchampion uses a completely different model, by the way: a database of all tasks keyed by uuid, with a separate list of un-sync'd operations. task undo is supported by that list of un-sync'd operations. It doesn't yet support task history (that's #2928).

[chaoticlonghair]

So, backlog.data can be safely deleted if I don't use taskd or task sync.

Do I understand correctly?