Cleaning the git history of a large repo

[Bisaloo]

This was discussed in the past with @ntorresd based on our experience with serofoi. It was later done by @sbfnk in EpiNow2 as well.

I think a blog post would be very helpful to:

• explain the issues with having a large repo
• explain the causes of a large repo, even if it doesn't seem to include large files in the current HEAD version
• present the recommended workflow:
  • make a backup
  • understand where the size issue comes from. Which files in history should be removed?
  • remove the files
  • push the new history
  • communicate the change and invite contributors to clone again
• present the gotchas:
  • can disrupt contributor flow
  • may or may not automatically close open PR. I'm not entirely clear on this one. I remember we had this issue with serofoi but it didn't seem to happen for EpiNow2 so maybe we did something different, or GitHub has changed something from their end.
  • ...

[chartgerink]

Coincidentally today, I found an application that aims to make cleaning the history of a repo more intuitive. I haven't tried it out, but might be of interest in this: https://retcon.app

[Bisaloo]

Thanks, this looks great!

I have edited the title to make it a little bit more specific: the intent here is potentially to talk about fixing git repos that became very large by rewriting history. I would be worth checking if this tool can do this kind of this because that's quite tedious at this time.