On Wed, Apr 13, 2016 at 01:08:29PM -0700, Linus Torvalds wrote: > On Tue, Apr 12, 2016 at 6:45 PM, Rich Felker <dal...@libc.org> wrote: > > > > Please pull these changes (regression fixes only) for arch/sh. They're > > based on 4.6-rc1 when I did them, but apply cleanly to 4.6-rc3 and > > build successfully. > > So I pulled this, but please don't do this: > > 16b02d711f40 Merge tag 'v4.6-rc1' > > there's no information in that merge commit why it would be needed, > and I cant' for the life of me see *why* it would be needed. > > If you cannot explain why a merge is necessary, you should not do the > merge. It's really that simple. > > So please > > - either just apply patches on top of your tree (no "let's merge > Linus' tree first") > > - or make your tree *start* at whatever base you want to use (ie > "let's check out v4.6-rc1, and apply patches on top of that base > commit")/ > > But do *not* start doing back-merges that aren't explained.
Sorry about that. My reason for starting with 4.6-rc1 was that I only tested the changes on it, not my previous (pre-merge) version. But I should have just rebased them on a clean branch from 4.6-rc1 rather than merging my own, right (your option 2 above)? > The back-merges make history harder to follow, and makes the graph > that gitk shows much messier. And _any_ commit that doesn't actually > explain why it is doing something is wrong, whether it's a merge or > not. > > Anyway, the pull is in my tree, and I'll push it out soon, so you > don't need to do anything for this one. This complaint was purely a > "going forward" issue. Thanks! Rich
