[Feature Request] ARC reset function.

[eik-dahms]

Sometimes user make mistakes and instead of tampering with the git history it would it is easier to "reset" the ARC. By that I mean - delete .git folder and .gitattributes file -> reopen ARC and reinitalizing git.

would be nice to have a function somewhere. Maybe in the history tab

[Brilator]

Do you really want to provide a button to delete all history and the remote connections?

[eik-dahms]

Yes I would prefer it maintenance wise. It can be a bit more hidden such that it is not easy to use + add a warning like you will delete all history etc. But I had it to do now several times by hand. Remote connections is not a problem. And yes you delete the history but most users don't use it properly anyways.

[Thyra]

It's not even just about remote connections; if the ARC has been pushed before, you'd need to either force-push or delete the remote copy or push to a new place and have a dead copy/duplicate lying around on the DataHUB.
I don't have a very good feeling about this feature 😅. Maybe there are cleaner ways of making error-fixing easier?

[eik-dahms]

OK this is just because can do a lot of stupid things + lfs tracking can be different for different datahubs. In our case I do not allow LFS tracked files outside of the dataset folders. because of the autotracking and sometimes lfs tracking for some users dificult to understand. This led to me deleting the .git + .gitattributes with them to reset everything. it happens mostly in the beginning and just resetting was the easiest way to keep frustration low on all sides.
I am not sure if you could even push to the old repo once this has been reset.

But yes maybe it is not a good idea after all and maybe documentation in the knowledge base where I can point to is sufficient. Ill try to do that next week then and we can close the issue if there are no more points.

[Thyra]

@eik-dahms "I do not allow LFS tracked files outside the dataset folders". This would be an important point to raise in https://github.com/orgs/nfdi4plants/discussions/30 as well and implies the bigger overall question of: should different datahubs have different LFS policies? Maybe that would be a good topic for a Data Steward circle or the matrix chat. Having the same policy everywhere would make things simpler and more transferrable for users but perhaps there are differences in technical setups where different policies make sense.

[eik-dahms]

I think this is not a question of should but rather of resources. Not everybody has a big S3 backend or in general even the space. Somewhere I wrote a comment or issue, which I can't find anymore, that I think it would be good that each datahub has e.g. policy.json document in its public available folder, which defines rules for the specific datahub. If it does not have this file "dataplant" standard settings should apply. This way each software working with arcs can apply these rules for the specific datahub. Having these rules would also make it easier to transform between datahubs.

But sure maybe there needs to be a discussion about this.

[JonasLukasczyk]

Lets include this in the discussion #506 where I recommend to use just the default rule where everything under datasets is tracked with lfs.