Hacker Newsnew | past | comments | ask | show | jobs | submit | alptony's commentslogin

From the "tmux-users" mailing list:

> Will pull requests on GitHub now be allowed as a means of contributing patches?

> No, patches still need to come by email to me [Nicholas Marriott] or the ML.

I wonder, what's the main reason behind it? What's wrong with Github PRs?


There are lots of big projects that use an email-based code review workflow: the gcc family (gdb, libstdc++, and gcc itself), git, Linux, Mercurial... You submit the patch to the mailing list and people comment on your patch inline. You submit updated patches as necessary until everything looks good and approved.

It's pretty low-tech, but like plain text itself, it's also very easy to participate. There's no setup involved other than an email client. And rewrites don't require you to add more commits, thus cluttering history; or to `git push --force`, thus destroying the previous version of the patch that people had commented on. It's also trivial to cherry-pick and rebase: just apply the patch. You can do this even with a VCS that doesn't have rebase or cherry-pick commands.

There are definite advantages to the email-based system.


Even if there are no problems with github PRs in themselves, the tmux developers will certainly have an established workflow for reviewing patches and applying them. The last thing they want to do while they're migrating from one hosting site to another is to upend their entirely functional code review workflow when that's not the thing that's broken and forcing the migration...


I don't know about tmux, but personally I find that it's pretty rare that you want to apply a submitted patch verbatim.

Generally there's a need to clean it up a bit and bang on it with some testing. I suppose that bigger projects can just bounce that back to the submitter to fix up and resubmit, but if a smaller project wants to encourage contributions the maintainers really need to do that.

There's also the point that projects with established mailing lists where patches are discussed probably don't want to move to github's web UI for those discussions.


I believe, and I might be very wrong, that tmux is developed in the OpenBSD CVS tree, and patches are then merged from there to Github. The patches would need to be applied to the code in CVS, so a Github pull-request would be useless.


FreeBSD has worked out a combined svn and github pull request workflow, although mayb esvn is easier to do this with than cvs.


I don't know what issues tmux has, but these are the reasons Linus also refuses to use Github PRs: https://github.com/torvalds/linux/pull/17


The comments section.


I prefer GitHub comments over inane mailing lists where you either get everything, or nothing. GitHub's issue notifications are opt-in per issue, and still come by email if you want.

People who only take patches by email are being jerks.


Consider applying for YC's Winter 2027 batch! Applications are open till November 2.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: