After the fix of Bug27765, we have a ability to see the string with the certain language if any though, it makes harder to write the consistent rules with matching family, style and fullname.
Currently it take effects only when the above of objects are specified in the pattern though, one needs to write possibly all of pattern matching on the above objects because the result only contains strings what namelang expects. i.e.
will returns the family name in Japanese if any, but style and fullname is still in English. so if a rule contains:
it may not be evaluated due to fail on matching with Gothic, but possibly ゴシック in the font pattern.
Proposed solution for this issue is, always to add a string in English with weak binding. that should helps.
Fixed in git c20ac78b01df3f0919352bba16b5b48b3b5d4d6d
Author: Akira TAGOH <firstname.lastname@example.org>
Date: Thu Nov 15 16:37:01 2012 +0900
Bug 57114 - regression on FcFontMatch with namelang
After 7587d1c99d9476b6dd4dbe523c0204da700fed8d applied, family,
style, and fullname is localized against current locale or lang
if any though, the string in other languages were dropped from
the pattern. this caused unexpected mismatch on the target="font"
This fix adds other strings at the end of the list.