Re: Font Problem with 23.0.92.1 - wrong font used

[Taylor Venable]

On Tue, Apr 14, 2009 at 12:49:48PM +0900, Kenichi Handa wrote:
> In article <address@hidden>, Taylor Venable <address@hidden> writes:
> 
> > > Ummm, weird.  Please define this function:
> > > 
> > > (defun font-file-at ()
> > >   (interactive)
> > >   (message "%s" (aref (query-font (font-at (point))) 1)))
> > > 
> > > and run it while putting cursor on "abc..." of `default',
> > > `mode-line-inactive', and `tooltip' lines in *Faces* buffer
> > > that is shown by M-x list-faces-display RET.
> 
> > First, when the Japanese font is present:
> 
> > default: /usr/X11R6/lib/X11/fonts/misc/7x13-ISO8859-1.pcf.gz
> > mode-line-inactive: [the same]
> > tooltip: [the same]
> 
> > Strangely enough, the font shown as the example for 'default' is not
> > the font used in the buffer display; again it appears to be the semi-
> > condensed variant, whereas the font used in the buffer display appears
> > to be the correct normal variant.
> 
> What is reported when you run font-file-at on a character
> displayed by that normal variant?

/usr/X11R6/lib/X11/fonts/misc/7x13.pcf.gz

In other words, it seems as if this is the font that should be used
for the 'default' face.

> And, the XLFD of 7x13-ISO8859-1.pcf.gz is:
>     -misc-fixed-medium-r-normal--13-120-75-75-c-70-iso8859-1
> Please run this and check if it is the font you mean by
> semicondensed variant?
> 
> % xfd -fn -misc-fixed-medium-r-normal--13-120-75-75-c-70-iso8859-1

No, this looks like what I want; in xfontsel it's specified by sWdth =
normal and avgWdth = 70.  What I've been calling the "semi-condensed
variant" is specified by sWdth = semicondensed and avgWdth = 60.

However, I can tell by just looking at it that when that font file
name is reported, the font being used is not the one with the XFLD
listed above and in my .emacs file.
 
> Also please check if the font shown by this is what you mean
> by normal variant:
> 
> % xfd -fn -misc-fixed-medium-r-normal--13-120-75-75-c-80-iso8859-1

This is too wide; the font that I've told Emacs to use as the default
has an avgWdth of 70, not 80.
 
> > If I do M-x customize-face default I notice that the value
> > has been set outside customize, and telling it to set for
> > the current session redisplays the buffer using the semi-
> > condensed face.
> 
> > When I run this function on 'bold' I get nil.  Finding a face that
> > looks like what I want and see in the buffer display, the function
> > reports the value as /usr/X11R6/lib/X11/fonts/misc/7x13.pcf.gz - but
> > the font for the 'bold-italic' face which also looks correct (there is
> > no italic slant variant for the -misc-fixed font I'm using) is
> > /usr/X11R6/lib/X11/fonts/misc/8x13B-ISO8859-1.pcf.gz

Is there meaning to the fact that the function returns nil for the
'bold' face?  Or perhaps it's an unrelated issue?
 
> > ========================================
> 
> > Without the Japanese font in the font path, I get these:
> 
> > default: /usr/X11R6/lib/X11/fonts/misc/7x13.pcf.gz
> > bold: /usr/X11R6/lib/X11/fonts/misc/8x13B-ISO8859-1.pcf.gz
> 
> I still can't reproduce that phenomenon; i.e. the existence
> of Japanese fonts affects the font selection.

This is on OpenBSD so maybe there's something weird in how X is setup;
I'll find a Linux box today to test on and see if there's any similar
or different behaviour there.

Thanks,

-- 
Taylor Christopher Venable
http://real.metasyntax.net:2357/