On Tue, Oct 03, 2017 at 04:10:12PM +0900, Junio C Hamano wrote:
> Jeff King <peff@xxxxxxxx> writes:
>
> > I'd prefer not to revert. I think setting any of the color config to
> > "always" in an on-disk file is basically a broken config. It was
> > exacerbated by 4c7f1819b, but it was already broken for scripts that
> > call "git log" or "git diff", or even just something as simple as piping
> > those programs on the command line.
>
> I actually disagree with that reasoning.
>
> We've promised that plumbing commands were safe to use in scripts,
> and 4c7f1819 ("make color.ui default to 'auto'", 2013-06-10) got it
> closer to breaking it (but not quite), and 136c8c8b finally broke
> it. Setting ui.color=always and shooting themselves in the foot by
> seeing ANSI escapes in "git log >file" output is totally user's
> choice. Breaking scripts that carefully chose to use plumbing,
> believing our earlier promise, and blaming user's ui.color=always
> does not sound quite like the same thing, at least to me.
I agree it's not quite the same thing, and I agree the problem was made
much worse by 4c7f1819b. But I still think color.ui=always is
inherently a foot-gun, and in either case it is the user that sets it
that is harmed (and they are the ones who have the power to fix it).
At any rate, I think you only need to agree with that reason to buy into
my "do nothing" option.
Option 2 would make everybody happy, I think (I'm working up a patch,
though note that it isn't quite trivial because of some abuses of the
color system in the test scripts).
-Peff
