cloud versioning: add gc support

iterative / dvc

🦉 Data Versioning and ML Experiments

https://dvc.org

Apache License 2.0

13.87k stars 1.19k forks source link

cloud versioning: add gc support #9360

Closed dberenbaum closed 7 months ago

dberenbaum commented 1 year ago

Add support for:

[ ] dvc gc --not-in-remote (see https://github.com/iterative/dvc/pull/9350)
[ ] dvc gc --cloud

daavoo commented 1 year ago

Is it possible to have this feature implemented? I believe it is essential to keep consistency on the remote side. Thanks in advance!

@manucalop would you expect DVC to "hard delete" the old versions of the files?

manucalop commented 1 year ago

Hi @daavoo ! Thanks for the quick reply :) I'd expect it to delete the live object, leaving its version history to be managed by the cloud-versioning system.

nullhack commented 9 months ago

I've been exploring DVC and I've also noticed this feature gap. Ideally, I'd prefer that the cloud storage's state mirrors the current state of the DVC repository. My expectation is that when using dvc gc -cw, it should remove the current version of any deleted file, aligning the cloud storage more closely with the repository's present status