On Mon, Feb 11, 2013 at 08:15:51PM -0800, Bryan O'Sullivan wrote:
I know you're fond of darcs, but it's well established by experience that we'll get many more contributions (both patches and problem reports) on github.
Have you got any evidence for that? I've just skimmed the "RFC: migrating to git" thread on glasgow-haskell-users@ in Jan 2011, and a handful of people gave some sort of indication that they might contribute (or contribute more) if GHC and its libraries switched to git. However, looking for them in the git commit logs (which admittedly isn't necessarily going to find all contributions from them, but it's the best I can easily do) for the almost-2-years since we migrated finds very few patches from them: Adam Wick <awick@galois.com> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019788.h... And just to show that you can't make everyone happy ... I'll put in an equal-and-opposite vote from Malcolm. GHC's use of darcs is a blocker for me even thinking about doing any work on GHC or the related libraries; my limited experience with darcs has been terrible, and the swearing I've heard about darcs makes me think my experience was not unique. Switching to git would make the chance that I'd do work on GHC nonzero. $ ./sync-all log --since "1 April 2011" | grep -i Wick $ Daniel Peebles <pumpkingod@gmail.com> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019761.h... I fully support this (especially if it lived on github), but we should probably sort the top contributors to GHC in the past year or so and consider their opinions on the matter in that order :) I certainly would not be on that list. A git(hub)-based workflow would however facilitate any minor contributions I might make (and I'd imagine those of many others). $ ./sync-all log --since "1 April 2011" | grep -i "Peebles\|pumpkin" <finds 4 patches, of which 3 are really one change split over 3 repos> $ ./sync-all log --before "1 April 2011" | grep -i "Peebles\|pumpkin" <finds 2 patches, related to a single change> Chris Dornan <chris@chrisdornan.com> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019772.h... I suspect a transition to Git would work out well in the long run and make the GHC sources more accessible. $ ./sync-all log --since "1 April 2011" | grep -i Dornan <finds 10 patches, but they're all in xhtml, which is maintained by Chris> Norman Ramsey <nr@cs.tufts.edu> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019777.h... I'd be thrilled to see GHC migrate to git, and I'd be much more likely to make new contributions to the back end. $ ./sync-all log --since "1 April 2011" | grep -i Ramsey <12 lines match, but all in hoopl, which is maintained by Norman> Trevor Elliott <trevor@galois.com> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019781.h... I am very interested in contributing to GHC, though the state of development with darcs makes me hesitate. A switch to git would make contribution to the project much easier. $ ./sync-all log --since "1 April 2011" | grep -i Trevor Author: Trevor Elliott <trevor@galois.com> $ I skipped 2 people, but I believe they both did internships at MSR and/or a SoC project involving working on GHC, so I don't think they necessarily tell us much about the VCS choice. David Peixotto <dmp@rice.edu> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019791.h... As an infrequent contributor I would welcome the move to git. I think the biggest advantage from my perspective would be enabling branches which I have avoided up to now because of the painful process I hear about from others. Max Bolingbroke <batterseapower@hotmail.com> http://www.haskell.org/pipermail/glasgow-haskell-users/2011-January/019753.h... Well, as a sometime-contributor I would certainly be happier hacking on GHC if it were git based. Thanks Ian