Eh, you really shouldn't do that for any kind of file that acts like a (an impromptu) database. This is how you get corruption. Especially when change information can be split across more than one file.

Sorry, what are you saying shouldn't be done? Backing up untracked/modified files in a bit repo? Or compressing the .git folder and backing it up as a unit?

> Backing up untracked/modified files in a bit repo?

This. It's best to do this in an atomic operation, such as a VSS style snapshot that then is consistent and done with no or paused operations on the files. Something like a zip is generally better because it takes less time on the file system than the upload process typically takes.

I see what you mean, but isn't this an issue with any filesystem backup tool? Or is there something about untracked files in a git workspace that is different, that I'm not seeing?