On Wed, Dec 4, 2024 at 8:54 AM Francesco Pretto <ceztko@xxxxxxxxx> wrote: > Ok, maybe now I get it, and what I missed is that prioritized > "objects" are not evaluated unless the ones before produced same > score. Is it possible that priorities in the documentation > (https://www.freedesktop.org/software/fontconfig/fontconfig-user.html#AEN21) > are not in sync with actual priorities in the code > (https://gitlab.freedesktop.org/fontconfig/fontconfig/-/blob/main/src/fcmatch.c?ref_type=heads#L323) They aren't the order of the priority though, Sure. Let me think about it. > ? If so the only thing I still don't understand is why :postcriptname > partial matched fonts with different styles disappear from the sorted > list returned by "fc-match -s", except from the one that was selected > with higher priority. If they were all present I would still be able > to re-filter the sorted list, looking for the desired style, but as it > is now it seems only one font per family can be returned (this is more > evident in my testbed [1]). That is the design. What you want would be "fc-match -a". > In general the situation is a bit unfortunate, because FC is a font > matching system with valuable qualities, but as it is now I can't > tweak it to express some simple (although unorthodox) queries. I > would RFE also this, but I think the "comma" normalization below has > much more priority for me, and let you think if in the future the > object priority list could be allowed to be altered so that my query > becomes possible. Yes, I'm pondering implementing that feature. > No, I haven't found actual fonts with a postscript name like that, but > many PDF(s) will still hint a postscript like name that has that comma > convention. This means that trying to locate a font using a :postscriptname > match will fail with FontConfig. PDF(s) usually have embedded fonts but > there are some legit cases where the font is not embedded and must be > located on the system. I see. > > However there are no such families of "Liberation" > > available. thus it will be marked as the lowest score. > > Ok. I guess that there's also no similarity/partial match for family name. It will more or less affect a score though, in a narrow sense, no. -- Akira TAGOH
