Git conflicts

emher
Level 3
Git conflicts

I just looked through the documentation on how to resolve conflicts,

https://doc.dataiku.com/dss/latest/collaboration/version-control.html#handling-conflicts

As a read the proposed work flow,

  1. Create a new branch B from the current branch A
  2. Push B to remote
  3. On the remote, merge A into B
  4. In DSS, switch back to A and drop local changes
  5. Pull

only the new branch B will be up-to-date (due to the merge in 3), so the steps 4-5 doesn't really make sense. Is it a bug in the documentation, e.g. should branch B be merged into A? Or is there another proposed workflow?

Working with many team member, we experience merge conflicts on a daily basis, so we are looking for the best possible workflow to resolve them.

 

0 Kudos
5 Replies
Andrey
Dataiker Alumni

Hi @emher , thanks for letting us know. In the step 3 the direction should be reversed: a merge should be done from branch B into branch A.

 

Regards

Andrey Avtomonov
R&D Engineer @ Dataiku
0 Kudos
emher
Level 3
Author

Hi @Andrey ,

thanks for the update. Is there a reason why you don't do it the other way around? That is,

  1. Create a new branch B from the current branch A
  2. Push B to remote
  3. On the remote, merge A into B
  4. Pull

This way, you avoid the switch-branch-and-drop-changes step.

0 Kudos
Andrey
Dataiker Alumni

@emher , Indeed this would give the same result, however the benefit of merging B into A is that the target branch never changes and the auxiliary branches are short-living and can be removed after the merge is done.

Andrey Avtomonov
R&D Engineer @ Dataiku
0 Kudos
tgb417

@emher

I'm starting to work with Remote git and collaborations.

It's been a bit over a year since you posted this original post. I'm wondering if you have worked out new and better approaches to collaboration on DSS and the use of remote git repositories.

--Tom
0 Kudos
emher
Level 3
Author

Unfortunately, I haven't made any progress in this regard. Working with git in Dataiku remains painfull. I think this is a point where Dataiku really needs to improve.