Nick Coghlan writes: > By contrast, proposals to switch from Mercurial to Git impose a > *massive* burden on contributors that don't already know git.
Let's not get carried away here. The *massive* burden is the moaning from git-haters (is there a 12-step program for them?) Agreed, learning any new VCS is a burden, and git seems harder than most -- but even RMS is swallowing git for Emacs, despite the fact that the "g" in "git" doesn't stand for "GNU". > with your GitHub ID, and the functional isomorphism [git <-> hg] I agree that there is an isomorphism, but the philosophical restrictions on hg functionality are quite annoying. I do things like "git reset" and "git commit --amend" a lot. I tend to commit before getting coffee, but I don't want that in the permanent record -- "git rebase --interactive" is a good buddy of mine. And so on. hg *deliberately* gets in the way of such workflows (although perhaps it's not as hard to "opt in" to the necessary features as it used to be). Nevertheless, I tend to agree with you that moving to Github now is a big move. I just think you should avoid the "git dox suck" argument. I'd also like to mention that in my opinion the network externalities argument is being misused. True, "everybody" has a github account, and even if they don't, their little sisters do. So what? There are big network externalities involved, but that doesn't necessarily mean that Bitbucket can't catch up, and most projects I know have branches hosted on both Bitbucket and Github (and often SourceForge or Launchpad as well). People who really prefer one or the other for practical reasons can usually use them without too much difficulty, wherever the "official" repo may be hosted. More likely to have a clear outcome, the main network externality *we* should be concerned with is *within* the Python ecosystem. *If* the "big" projects whose core members tend not to hang out here so much (NumPy, SciPy, Twisted, Django, ...) are vastly more likely to be found on Bitbucket than Github (or vice versa), I think that's potentially much more important than the little sisters with github accounts. I also agree with you that the facts that Mercurial is a Python application, and I guess so is most of Github, are important. But again, let's not get carried away. "Although practicality beats purity" applies here. The Github features are very attractive; we need to look at how useful they will be to contributors before deciding that the warm fuzzy Python community is more important. Finally, Guido is right: Github, and to a somewhat lesser extent Bitbucket and Google Code have gotten code hosting right, compared to the *forges. The people who maintain infrastructure for Python are the kind of contributor who would probably spend more time on reviewing and mentoring and release engineering if they weren't maintaining infrastructure as far as I can see. If the infrastructure maintenance can be delegated (it's not clear to me that it can), that would be a big factor. _______________________________________________ Python-Dev mailing list Python-Dev@python.org https://mail.python.org/mailman/listinfo/python-dev Unsubscribe: https://mail.python.org/mailman/options/python-dev/archive%40mail-archive.com
