there are to important points to keep in mind: (1) we want to trust the main stable branch (whatever that is) as much as possible
This implies that arbitrary commits should get reviewed. However simple patches can be reviewed most efficiently by "batch processing". (2) we want new users to start contributing/testing as early as possible, we want to make it easy for people to collaboratively work on patches or mutliple ways to solve the same issue. And this can be done most easily by allowing write access to public. IMPORTANT: This implies that people know what they are doing - and that they may run the risk fetching malicious code this way. So whatever you get from there should be reviewed by the comitters themselves We don't want to give arbitrary people commit rights on nixos.org, but github also offers gists. So does it make sense to create official nixos/nixpkgs gists people can push their ideas to ? Everybody can grant him/herself write access easily. We could create a branch "trivial patches" which could be merged into the main branch on a regular basis by batch processing done by core members we trust (which may push to nixos.org) This way no commits get lost - and some reviewing and commenting can take place easily. Some features seem to be disabled on gists such as commenting :( Then it would be possible to write a 5 line guide about how to become a nixos contributor: checkout the gist, make sure nobody pushed a bad master HEAD, commit your stuff and push (on a different branch which might be called trivial-patches) I'd even allow rewriting history on that repository so that it can be cleaned up easily. Most patches on the trivial-patches branch would go away after rebasing on master after those patches got cherry-picked. Its not perfect - but it might work well for many cases. pull requests are fine for more complicated patches because commenting can be done etc - but I agree on them being overkill for trivial version updates and the like. Of course top-git branches could be used if appropriate on gist repos, too Marc Weber _______________________________________________ nix-dev mailing list nix-dev@lists.science.uu.nl http://lists.science.uu.nl/mailman/listinfo/nix-dev