Are face specifications "frame local"?

Are face specifications "frame local"?

[Alan Mackenzie]

Hello, Emacs.

I've been getting confused a bit the last couple of days by newly
created faces appearing to be "frame local".

The faces are created in old software by make-face and copy-face (yes, I
know!) and appear to apply only to existing frames, not frames created
later.  For example, if I have a buffer on a frame displaying as magenta
text with face font-lock-label-face (which doesn't exist in Emacs), and
I do C-x 5 2, creating a new frame, what is magenta on the initial frame
becomes neutral on the new frame, even though it still has the "same"
face.

I haven't find any helpful description of "frame local" faces in the
Elisp manual, and their relationship to "global" faces.  Just that some
functions on, for example, page "Attribute Functions" describe optional
FRAME parameters and say these indicate the use of the "frame local"
values.

I get the impression that "frame local" faces are a dubious feature, and
that the new(?) macro defface is trying to phase these out.  I can't see
any use case for "frame local" faces.

What do people say?

-- 
Alan Mackenzie (Nuremberg, Germany).

Re: Are face specifications "frame local"?

[Eli Zaretskii]

> Date: Tue, 13 Sep 2022 17:49:05 +0000
> From: Alan Mackenzie <acm@muc.de>
> 
> I haven't find any helpful description of "frame local" faces in the
> Elisp manual, and their relationship to "global" faces.  Just that some
> functions on, for example, page "Attribute Functions" describe optional
> FRAME parameters and say these indicate the use of the "frame local"
> values.
> 
> I get the impression that "frame local" faces are a dubious feature, and
> that the new(?) macro defface is trying to phase these out.  I can't see
> any use case for "frame local" faces.

I don't know where you get this.  Faces are frame-local in Emacs,
always were.

If there's some real-life problem you need to solve, let's hear the
details and see some code, and then let's help you find the solution.
There's no point ion discussing this on philosophical level.

So: what is the problem you are trying to solve?