Eli Zaretskii <eliz@gnu.org> writes:

>> > (progn
>> >  (setq-local text-scale-remap-header-line-face t)
>> >  (setq-local header-line-format (concat
>> >                                  (propertize "foo" 'face 'fixed-pitch) " "
>> >                                  (propertize "bar" 'face 'variable-pitch))))
>> >
>> > They also scale.
>>
>> They don't scale obn my system, which is what I'd expect.  Maybe these
>> faces don't specify the font on some systems, but the principle is
>> nevertheless valid.
>
> No, that's because your patch is not yet in, so setting
> text-scale-remap-header-line-face has no effect.  Sorry, I didn't
> realize you haven't yet committed those changes.
>
> In any case, the general point is valid: it is incorrect to assume
> that every face used anywhere will scale with the buffer text.

What do you think of the attached patch?  It is along the lines of
what I suggested in a separate email, where I wrote:

> So looking at this, I'm not sure we should even say: "This will also
> affect any face that inherits from it."  It risks confusing more than
> helping, IMHO.
>
> Nor am I so sure we should talk about the 'header-line' face.
>
> I think we should say something closer to what Martin suggested, say:
>
>   *** text-scale-mode can now scale the header line.
>   When the new buffer local variable 'text-scale-remap-header-line-face'
>   is non-nil, 'text-scale-adjust' will also scale the text in the header
>   line when displaying that buffer.
>
>   This is useful for major modes that arrange their display in a tabular
>   form below the header-line.  It is therefore enabled by default in
>   'tabulated-list-mode' and its derived modes.
>
>   Note that text scaling never affects text using a face where the
>   :height attribute is set to an absolute value.
>
> And then adjust the docstring accordingly.

Do you think this is horribly wrong, or does it make sense?

Thanks in advance.