Re: 23.0.60; SEGV in font_score()

[James Cloos]

OK.  That was exactly what I needed.

The non-symbol family was an int.  300.

And I found this in fc-list:

:; fc-list 300 file family
/home/cloos/.fonts/Urdu/Nafees_Pakistani_Naskh.ttf: 300,Nafees Pakistani Naskh
/home/cloos/.fonts/Urdu/nafees_naskh.ttf: 300,Nafees Naskh
:; fc-cat ~/.fonts/Urdu|grep 300
"Nafees_Pakistani_Naskh.ttf" 0 "300,Nafees Pakistani Naskh:familylang=en,en
 :style=Regular:stylelang=en:fullname=Nafees Pakistani Naskh:fullnamelang=en
 :slant=0:weight=80:width=100:foundry=unknown:index=0:outline=True:scalable=True
 :charset=<elided>:lang=:fontversion=131072:capability=otlayout\\:arab
 :fontformat=TrueType:decorative=False"
"nafees_naskh.ttf" 0 "300,Nafees Naskh:familylang=en,en:style=Regular
 :stylelang=en:fullname=Nafees Naskh:fullnamelang=en:slant=0:weight=80
 :width=100:foundry=unknown:index=0:outline=True:scalable=True
 :charset=<elided>:lang=:fontversion=131072:capability=otlayout\\:arab
 :fontformat=TrueType:decorative=False"

Those two fonts came from www.crulp.org and, as you can guess from where
I saved them, target Urdu.  Based on the filesystem dates, they've been
in my .fonts for more than 3 years.

I couldn't find a link to either of them on their current site, so I
grabbed their current fonts and moved those two out of the way.

In short, Emacs stored the family name as an int instead of as a symbol
if it happened to be composed on only ascii digits.

I suspect the name 300 is a bug in the fonts, but you'll probably have
to defend against such possibilities anyway.

-JimC
-- 
James Cloos <address@hidden>         OpenPGP: 1024D/ED7DAEA6