On Tue, Nov 8, 2022 at 6:24 AM Robin A. Meade <robin.a.meade(a)gmail.com> wrote:
Hi Fonts SIG,
Currently there is poor discoverability of what fonts provide a glyph for a particular a
It was recently asked  what font package in Fedora is needed to display U+130DF .
## Method 1: Download all font packages and use fc-list
One way to determine this would be to install every font package and then issue the
following `fc-list` invocation: `fc-list :charset=130DF`:
sudo dnf repoquery --available --whatprovides 'font(*)' | xargs sudo dnf install
Aside: Is there a better way to list all font packages than this:
sudo dnf repoquery --whatprovides 'font(*)'
Well, a dependency tag of "font(*)" was too much and problematic to
decide what should be installed by default according to the language
coverage. so it has been moved to langpacks now. you can simply try
"dnf install langpacks*" or "dnf install @fonts" assuming that
packagers are updating comps. If you don't rely on them, you could
try "dnf install *-fonts" of course.
## Method 2: Pipe Dream: Declare code points as PROVIDES in the RPM
It would be cool if there was somehow a way to declare in the RPM spec of the font RPM
what code points this font has glyphs for. Then something like `sudo dnf whatprovdes
glyph(130DF)` could be used to discover the fonts that have a glyph for U+130DF.
Probably not feasible, but throwing it out there.
No, it isn't realistic and too much to cover all the glyph coverage in
a font. we could add something per the Unicode code blocks instead,
which looks like "lang" property though, it doesn't guarantee how much
it actually covers (like what "lang" currently presents). In that
sense, it may actually be a duplicate feature of "lang". Or it could
be flagging only when it covers all in a certain block. though I'm not
sure how much the fonts satisfy them. in either way, I don't think it
would really help you to find out "U+130DF".
## Method 3: A Fedora website: Fedora Font Showcase
A website could be designed that showcases the fonts Fedora has packaged.
It could include a **search by code point** feature.
A database containing the relationship between font packages and code points could be
created using the technique of Method 1.
This somehow sounds realistic to me. I'm not sure how to sync up
databases against package updates though, that way may be worth taking