Re: [PATCH 1/2] color: downgrade "always" to "auto" only for on-disk configuration

[Jonathan Nieder <jrnieder@gmail.com>]

Hi,

Junio C Hamano wrote:
> Jeff King <peff@peff.net> writes:
>> On Sat, Oct 14, 2017 at 12:01:46PM +0900, Junio C Hamano wrote:

>>> True.  Let's see what others think.  I know Jonathan is running
>>> the fork at $work with "downgrade always to auto" patches, and while
>>> I think both approaches would probably work well in practice, I have
>>> preference for this "harder but right" approach, so I'd want to see
>>> different views discussed on the list before we decide.
>>
>> 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:

Which approach is "harder but right" / "peel back"?

I agree with the goal of making color.ui=always a synonym for auto in
file-based config.  Peff found some problems with the warning patch
(scripted commands produce too many warnings), which are not an issue
for $dayjob but may be for upstream, so I see the value of holding off
on the warning for now.

I'm also fine with "revert the proximate cause of the latest
complaints" as a stepping stone toward making color.ui=always a
synonym for auto in file-based config in a later release.

Another issue is diff-files paying attention to this configuration.
If I'm reading Documentation/config.txt correctly, that was simply a
bug.  diff-files and diff-index are never supposed to use color,
regardless of configuration.

I'm fine with "revert the proximate cause of the latest complaints" as
a stepping stone toward fixing that, too. :)

Sorry I don't have more detailed advice.  I was planning to look more
closely at how these features evolved over time and haven't had enough
time for it yet.

Jonathan