Re: Adequacy of min-colors thresholds

On Tuesday, July 14th, 2026 at 6:25 AM, Eli Zaretskii <[email protected]> wrote:

> > Date: Mon, 13 Jul 2026 17:44:17 +0000
> > From: Heime <[email protected]>
> > Cc: [email protected]
> >
> > On Tuesday, July 14th, 2026 at 1:33 AM, Eli Zaretskii <[email protected]> wr=
ote:
> >
> > > > Date: Mon, 13 Jul 2026 12:52:24 +0000
> > > > From: Heime <[email protected]>
> > > > Cc: [email protected]
> > > >
> > > > > The rule of thumb is: unless you must use very fine differences
> > > > > between colors, don't use min-colors at all.  Emacs is quite capa=
ble
> > > > > of choosing the best match to "darkseagreen2" even if a color by =
that
> > > > > name is not directly supported.
> > > >
> > > > I'm aware of the capability you mention.  Nevertheless, what values
> > > > would you suggest to test for min-colors?
> > >
> > > To test? all of them, of course.  Or at least as many as you can.  Yo=
u
> > > should test on 8-color terminal, on 16-color terminal, on 256-color
> > > terminal, and on TrueColor terminal.
> >
> > Would that be (min-colors 256) on 256-color terminal.  How about TrueCo=
lor?
>=20
> No, that's not what I meant.  I meant to define the color without
> min-colors at all, and _then_ test with terminals (or terminal modes)
> which support different numbers of colors

Examples show use of min-colors, but stop at 88.  Would it not be a good
idea to describe your suggestion in the manual to define colours without
min-colors?
=20
> > And what is a (min-colors 88) exactly?
>=20
> Any terminal that supports at least 88 colors.

Why the chosen 88?  Is that some special terminal support?