bug#73862: [PATCH] Add `header-line-active` and `header-line-inactive` faces.

[Eli Zaretskii <eliz@gnu.org>]

> From: Aaron Jensen <aaronjensen@gmail.com>
> Date: Tue, 3 Dec 2024 22:30:58 -0800

Aaron, it would have been more useful to CC Trevor, who is the author
of that changeset.  I've added him now.

> I'm able to reproduce this with the mode-line as well, so it appears that there may be a bug there too in the
> code that was copied to implement the active/inactive faces in the header line. It's entirely possible that
> there's a bug in nano-modeline, but it seems suspect that code was added to consider windows and now this
> issue triggers when new windows are created (and possibly when selection changes, I haven't eliminated
> whether or not that's a factor yet).

You were able to reproduce what?  I don't think you posted a recipe to
reproduce the problem.  Please do, if at all possible, preferably
starting from "emacs -Q".

> From: Eshel Yaron <me@eshelyaron.com>
> Cc: Trevor Murphy <trevor.m.murphy@gmail.com>
> Date: Wed, 04 Dec 2024 10:47:18 +0100
> 
> >     Add new `header-line-active' and `header-line-inactive' faces
> >     
> >     This is all intended to parallel the 'mode-line-active' and
> >     'mode-line-inactive' distinction.
> [...]
> 
> This seems to introduce a regression, consider the following recipe:
> 
> 1. emacs -Q
> 2. In the scratch buffer, evaluate:
>    (setq header-line-format "foobar")
>    (face-remap-add-relative 'header-line 'highlight)

Aren't you supposed to remap the two new faces instead of
'header-line'?

> 3. Type C-x C-M-= or something similar to force updating the header
>    line.  The header line in the scratch buffer now shows "foobar" and
>    uses the highlight face, as expected
> 4. Type C-x 4 b new RET to switch to another buffer in another window
> 5. In the new buffer evaluate (setq header-line-format "foobar")
> 6. Observe that the header line in the new buffer is also using the
>    highlight face.  That's unexpected!
> 7. Type C-x C-M-= while the new buffer is current
> 8. Observe that the header lines in both windows no longer have the
>    highlight face.  That's unexpected!
> 
> Before, remapping the header-line face with face-remap-add-relative
> would only affect the current buffer, as expected.  Now it seems like
> the face remapping "leaks" between buffers/windows somehow...

Trevor, could you please look into this?