I think if the committer does the squashing they should keep the author. I tried an experiment where I took a patching having multiple commits and squashed them into another branch. This then seems to work fine:
git commit --author "Some Name <[email protected]>" -m "This commit has lots of changes" It's definitely nice having a single commit instead of having all the intermediate commits interspersed. This would be a good policy to follow. On Thu, Jan 16, 2014 at 5:03 PM, Jakob Homan <[email protected]> wrote: > Yeah, that's the same thing I found as well and the process I mentioned > above. It'll certainly work. Chris and I had talked about some script to > automate the process, since the whole point is to make it easy to give > credit to the contributors, who may or may not be familiar with the patch > process. > > I like the process, just wish it could do the squash automatically. > > I'm not sure if the committer doing the squashing would still keep the > author? Maybe. > > > On Thu, Jan 16, 2014 at 4:37 PM, Matthew Hayes < > [email protected]> wrote: > > > What if the person creating the patch uses a throwaway branch like below > if > > there are multiple commits? Otherwise the one committing the patch could > > squash themselves but they'd have to write the commit message. > > > > > > > http://stackoverflow.com/questions/616556/how-do-you-squash-commits-into-one-patch-with-git-format-patch > > > > > > On Thu, Jan 16, 2014 at 3:28 PM, Jakob Homan <[email protected]> wrote: > > > > > We haven't actually done this quite yet for one reason. It turns out > > that > > > when you have multiple commits, this method replays those commits > against > > > the repo you're patching, ie your intermediate commits show up too. > This > > > may or may not be what people want; generally I just want the final > > version > > > of the patch to be applied. > > > > > > I looked around and couldn't find a way to automatically squash the > > commits > > > as part of this process, so if one wants to do that squashing, it must > be > > > done manually. > > > > > > Otherwise, and for single commit patches, this works well. > > > -jg > > > > > > > > > > > > On Thu, Jan 16, 2014 at 3:00 PM, Matthew Hayes < > > > [email protected]> wrote: > > > > > > > I wanted to share a link to this blog post that Jakob shared on the > > Samza > > > > dev mailing list: > > > > > > > > > > http://ariejan.net/2009/10/26/how-to-create-and-apply-a-patch-with-git/ > > > > > > > > Quoting Jakob: > > > > > > > > > Two benefits over the current, svn-style patches: > > > > > 1) The patches include the author, who is credit in the commit log > > > (I've > > > > > been doing this manually, but it's a pain) > > > > > 2) The patch application and commit happens as a single step, > helping > > > to > > > > > avoid dunce-level errors like mine from yesterday. > > > > > > > > > > Thoughts on this? Josh is this the practice you all follow in > Apache > > > > Crunch? > > > > > > > > If people agree we should add these instructions to the website for > > those > > > > who want to contribute. > > > > > > > > -Matt > > > > > > > > > >
