bug#47047: [External] : Re: bug#47047: 28.0.50; 'help-key-binding' face's background sub-optimal

[Stefan Kangas]

Drew Adams <drew.adams@oracle.com> writes:

> What do you mean by "faces defined"?  Do you mean new
> faces added by the theme, or existing faces redefined
> by the theme, or something else?
>
> Existing faces that are not redefined are nevertheless
> covered (still defined), I'd think.

I mean that the theme makes no special effort to support various modes.
So you might have a theme with a background color that is very close to
the foreground of a face defined in some package, leaving things
completely unreadable.  I've seen it many, many times.

> What inconsistent results have you identified?  What
> kinds of inconsistency?

I mean, you can literally just open up any special-mode and have a look.
For example, with `light-blue-theme', the eshell prompt is all but
unreadable, and `eshell-prompt' is not defined in the theme itself.

If you don't like that example, try some other modes like Gnus, notmuch,
mu4e, ERC, etc.  Whatever you like.  You will find problems, I promise,
and you won't even have to dig deep.

There is just no way to theme Emacs by just throwing some 25 odd face
definitions in a theme and call it a day.  Take a look at
e.g. manoj-dark.el for something a bit more realistic.

> There's plenty of code that's part of Emacs that hasn't
> been changed an iota over the years, but still works.
> Does that mean that it's necessarily unmaintained or
> incomplete?  Certainly any code could be enhanced, but
> that's something different, and not a requirement for
> its continuation.

I'm talking about themes that are very old, unmaintained and not very
complete.

In any case, I don't think we are going to get anywhere by just
exchanging emails in this already closed bug report.  I hope I made my
reasoning somewhat more clear at least.