On Mon, Oct 6, 2014 at 11:27 AM, Andreas Abel wrote: > This is also the thing that worries most about arc: Squashing > commits. Splitting commits into > > * things that only do whitespace changes > * things that only add comments > * things that only refactor > * things that actually introduce a semantic change > > *is* very valuable also for the efficency of the review process.
Having separate commits also increases the chances of finding the faulty diff with git-bisect. Also, reviewing is usually easier with split commits, and it's simpler to provide descriptive per-commit explanations within each commit message. That being said, temp/fixup/backup commits should be squashed prior to submitting a commit/patch series, and I suppose that's the rationale behind arc's behavior. > On 05.10.2014 19:13, Tuncer Ayaz wrote: > > > > On Sun, Oct 5, 2014 at 4:32 PM, Ben Gamari wrote: > > > > > > 5.a. I reflected on the mild shock of seeing that `arc` had > > > squashed my carefully crafted patch set into a single > > > commit. This still bothers me to this day. > > > > > > I second 5.a, but does it have to be this way, or can arc be > > instructed to not squash commits? _______________________________________________ ghc-devs mailing list [email protected] http://www.haskell.org/mailman/listinfo/ghc-devs
