Re: [Emacs-diffs] emacs-25 1c58fa1: Fix variable-pitch font on MS-Windows

[Eli Zaretskii <eliz@gnu.org>]

> From: Stefan Monnier <monnier@IRO.UMontreal.CA>
> Cc: Eli Zaretskii <eliz@gnu.org>
> Date: Sat, 30 Apr 2016 09:48:32 -0400
> 
> > -  '((t :family "Sans Serif"))
> > +  '((((type w32))
> > +     ;; This is a kludgey workaround for an issue discussed in
> > +     ;; http://lists.gnu.org/archive/html/emacs-devel/2016-04/msg00746.html.
> > +     :font "-outline-Arial-normal-normal-normal-sans-*-*-*-*-p-*-iso8859-1")
> > +    (t :family "Sans Serif"))
> 
> This is really curious.  Why does Emacs pick a bold-italic face when we
> don't specify bold (nor italic)?

I have no idea.  My best guess is that bold-italic font variant is the
first one that Emacs finds, but I have no evidence to support that
guess.

One thing is clear: Emacs doesn't know that the resulting face is
bold-italic, because manually forcing the corresponding attributes to
be "normal"/"roman" doesn't fix this.  (Specifying those attributes
explicitly was a solution I tried first, because it seems like less
ugly.)

> I understand that the above defface doesn't explicitly specify ":slanted
> nil", but it's still very puzzling.

I very much agree.  Unfortunately, font selection logic is something I
have only a very vague idea about.  The change which apparently caused
this as side effect was for a similar (and more serious) problem, and
the solution I found was just a heuristic, by watching the fonts that
were examined and rejected.