On Tue, Oct 17, 2017 at 03:26:25PM +0900, Junio C Hamano wrote: > Junio C Hamano <gitster@xxxxxxxxx> writes: > > > Jeff King <peff@xxxxxxxx> writes: > > > >> After pondering over it, I have a slight preference for that, too. But > >> I'm also happy to hear more input. > > > > OK, so it seems we both have slight preference for the "peel back" > > approach. Adding Jonathan to Cc: > > It was a bit more painful than necessary to make sure I have > something that can be merged for 2.14.x maintenance track, but I > think the topic is now in a reasonable shape, and I've merged it to > 'next'. On the first-parent chain from 'master' to 'pu', the merge > of this topic is the very first one, and after reading it over once > again, I think this is OK. Hmm. I think you would just want the top two commits for maint-2.14 (reverting 136c8c8b8f and fixing up git-tag to check color config). But of course you can't do a partial merge because they come on top of the other dead-end/revert pair. You'd have to cherry-pick (and even then fix up a few bits, like adding in the "add -p" test). Though if we take all of jk/ui-color-always-to-auto-maint, and then do the whole reversion on top of that, I think that should work. It just doesn't look like that topic ever made it to "maint" (I see mention of a jk/ref-filter-colors-fix-maint in the log for master, but there's no such branch). I started to prepare a patch directly on v2.14.2 just to see what it would look like. The first one (the revert) is fine, but we then have to fixup tag and for-each-ref. And since they don't have --color added by the dead-end fixups, the tests get harder... -Peff
