mathog wrote:
[...] Is there currently some way to get fontconfig to reveal how good a match was found? If not, could it please be enhanced to return some sort of quality score? For instance "10" for
fontconfig uses a score internally in the matching stage, but it's (1) not stable, (2) has only relative meaning and (3) has nothing to do with font metrics, as it only covers the properties available in the fontconfig cache.
In your case, a good match would be one that is 100% metric-compatible. However, this is not possible to determine with the data available to fontconfig; it would imply that the metrics data for all missing fonts was somehow available.
The most strict implementation would be one that requires family name equality, but in this case you'd miss a number of good aliases.
A not as strict implementation would be based on Behdad's suggestion: http://thread.gmane.org/gmane.comp.fonts.fontconfig/4251
Another possible method is to use a restricted fontconfig configuration that you know only maps to metric-compatible aliases. With the exception of a few configuration files that are in fontconfig's default set (but are probably not guaranteed to remain in it forever), you'd have to write that yourself (and accept the fact that with it you ignore any configuration done by users/admins etc).
Maybe you could also check bounding boxes of the text you render (if you know them for sure) and squeeze your fonts with <matrix> elements.
In any case, I'd be careful not to overengineer stuff. Raimund -- Worringer Str 31 Duesseldorf 40211 DE home: <rs@xxxxxxxx> +49-179-2981632 icq 16845346 work: <rs@xxxxxxxxxxxxxxx> _______________________________________________ Fontconfig mailing list Fontconfig@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/fontconfig
