Enzo-3.0 repository cleanup

[Nathan Goldbaum]

Hi all,

Several months ago there was a merged pull request in the enzo-3.0 repository that unintentionally included a set of "bad" changes that substantially increased the size of the repository.

We would like to fix this.

I've gone ahead and rewritten history in my repository to include a version that doesn't descend from the "bad" changeset:

https://bitbucket.org/ngoldbaum/enzo-3.0/commits/branch/enzo-dev

Specifically, I'm talking about the commits between 60e3a9c and 32bb8c5

The net effect is to rewrite history for Britton and John's most recent merged PR and my PR that was merged this morning that merges with work in the enzo-dev repository.

The reason I'm writing is that we will need to strip public history. Anyone who is doing work in the enzo-3.0 repostiory will also need to ensure that their work is not based on the "bad" changeset. If you're concerned that your work will need to be redone, we will likely be able to extract your work using a rebase to prevent too much annoying repetition of work.

I'd like to strip the repository soon but will give this a day or two so people can raise objections by replying to this e-mail.

The alternate choice is to not rewrite public history at the cost of having a very heavy-weight repository on disk. We would also be much closer to the 2 gigabyte repository size limit that bitbucket imposes.

-Nathan

[Britton Smith]

Hi Nathan,

Thank you for sorting this all out.  I am 100% in favor of extracting these changes from the public history, especially since you have found a straightforward way to do so.  Consider me +1 on this.

Britton

--
You received this message because you are subscribed to the Google Groups "enzo-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email to enzo-dev+u...@googlegroups.com.
To post to this group, send email to enzo...@googlegroups.com.
Visit this group at http://groups.google.com/group/enzo-dev.
For more options, visit https://groups.google.com/d/optout.

[John Regan]

Also fine with me.

+1

[Tom Abel]

+1 

So nice to imagine it clean …

T

[James Larrue-Baulch]

Hi Nathan,

Do you have any idea yet about when this will happen?

James

[Nathan Goldbaum]

Hi James,

Unfortunately the bitbucket engineers I've talked to have not been responsive about this. I've brought it up several times on IRC, and they've said they'd look at it, but so far nothing.

An alternative would be to open a new repository and just abandon the old one?

I don't really know what the best course of action is, unfortunately.

Nathan

[Nathan Goldbaum]

I just tried again, and I managed to get Kaleb Elwert's attention, and the bad commits have now been stripped.

I'm going to push all the "good" commits among the ones that were stripped back up now.

[Nathan Goldbaum]

OK, I *think* everything should be back to normal now.

Please let me know if we're missing any pull requests. I had to rebase John Regan's particle splitting PR. In addition, any changeset that was a successor of the "bad" changeset now has a different commit hash. Everyone will need to re-clone the repository, and any open work that needs to be pull requested will need to be redone using the new public history.

It's likely that a lot of things will be automatable using mercurial's history rewriting capabilities. If anyone has a specific task they need done and don't want to do it manually, let me know and I'll likely be able to help you out. Just push the commits to your public fork of enzo-3.0 and tell me what history rewriting needs to happen.

Given that the "bad" changes were public for a very long time, I would like to request that pull request reviewers in the enzo/enzo-3.0 repository be *very* careful to make sure only the intended changes make it into the repository so we don't need to repeat this exercise.

-Nathan

[John Wise]

This is great, Nathan. Thanks for taking care of this!

John

> <james....@diopolis.com <mailto:james....@diopolis.com>>

> <mailto:enzo-dev+u...@googlegroups.com>.

> To post to this group, send email to enzo...@googlegroups.com

> <mailto:enzo...@googlegroups.com>.

> Visit this group at http://groups.google.com/group/enzo-dev.
> For more options, visit https://groups.google.com/d/optout.

--

John Wise
Assistant Professor of Physics
Center for Relativistic Astrophysics, Georgia Tech
http://cosmo.gatech.edu

[John Regan]

Yes great work Nathan! That last bit is probably good advice as well. I must admit I never checked PRs for extra files before.

I have a few forks of 3.0 on the go so I might ping you at some stage for assistance!

Cheers,

John