Large features are typically worked on by multiple authors and attribution needs to be preserved.
Multiple commits by the same author that don't carry meaning to anyone else and/or occur to address PR review comments should be combined. This consistently holds for the single author commit PR review as it does for the multi-author feature work. On Mon, Nov 2, 2015 at 3:31 PM, Pramod Immaneni <[email protected]> wrote: > If you are going to allow multiple commits in a pull request when there are > multiple authors I don't see why to have the restriction to merge always, > even after pull request has been submitted, to a single commit when there > is one author. I just find it very contradictory that history is being > suppressed after the code has come into the open, by virtue of it being > submitted in a pull request, when the whole point of version control is the > exact opposite. > > On Mon, Nov 2, 2015 at 3:22 PM, Vlad Rozov <[email protected]> > wrote: > > > There is no difference whether pull request is created by committer or > non > > committer, both need to create fork as described in > > http://apex.incubator.apache.org/contributing.html and squash commits as > > described in item #7. Only committer is able to merge from the fork to > the > > ASF master repository and git will preserve the author (or authors if > there > > are multiple commits from different authors). > > > > The recommendation to squash commit is not much different from other ASF > > projects commit policy where contributor submits single patch that is > later > > merged by a committer. Committer is not responsible for squashing > commits, > > this should be done by the contributor. > > > > Thank you, > > > > Vlad > > > > > > On 11/2/15 14:19, Justin Mclean wrote: > > > >> Hi, > >> > >> You may want to be careful if changes were made by multiple authors or > on > >> behalf of someone else (i.e. pull request by a non committer), > especially > >> if any of them have not signed an ICLA. IP provenance needs to be > >> preservered. In those cases I’d suggest not squashing commits. > >> > >> Justin > >> > > > > >
