bug#41544: 26.3; Possible incorrect results from color-distance

[Eli Zaretskii]

> From: Lars Ingebrigtsen <larsi@gnus.org>
> Cc: mattiase@acm.org,  simenheg@runbox.com,  tom@tromey.com,
>   41544@debbugs.gnu.org
> Date: Tue, 18 Aug 2020 16:10:44 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > How do they make things better, and what things, exactly?  My
> > recollection is that this was about consistency, not correctness.
> 
> According to the screen shots, they compute correct colours (based on
> fixing the luminance formulas, if I remember correctly (it's been a
> couple weeks since I read this thread)).

My bitter experience with handling colors in Emacs is that changes
frequently fix some use cases and botch others.  So I'd prefer to make
such changes only where the current results are completely
unacceptable (and I'd be surprised if we have such cases nowadays).

> > I don't think defcustom is the proper way of handling differences in
> > behavior at this low level.
> 
> Different ways of computing colours?  That's not very low level.

I think it is.  Think about what you'd say in the doc string of such a
defcustom.  "If non-nil, handle colors correctly, otherwise handle
them incorrectly"?