After many years on Bitbucket, today I moved coverage.py development to GitHub. I’ve long wanted to use Git instead of Mercurial, but I didn’t want to lose the issues that were on Bitbucket.
I thought GitHub issues were the simplest possible bug tracker. But I realized that Bitbucket’s were even simpler: they have no labels. So I bit the bullet, added some features to the Bitbucket issue migration tool, and switched over.
There had been a GitHub mirror of the code, but I started a new repo so that the issues transferred from Bitbucket could keep their old numbers. The old mirror is still around, because there’s a handful of issues and pull requests there that I didn’t want to lose.
This move is also a kind of way to declare some pull request bankruptcy. I’ve embodied the bad habit of letting pull requests languish, sometimes for years. I shouldn’t let that happen, but I can’t promise that I will do better. I’ll try. At least Git will give me a more familiar environment in which to work with them, which I hope will help.
Another side-effect of making a new repo is that I have hardly any stars yet...
I picked Mercurial just before it became obvious that Git would be the most used (a sort of Betamax decision). I'm trying to see an advantage to Git which would motivate me put up with its messy user interface (https://xkcd.com/1597/).
Add a comment: