Am 17.07.2018 um 22:47 schrieb Aaron
Hill:
On
2018-07-17 13:37, Urs Liska wrote:
Oh my, then again I get:
Aachen Std
['Aachen Std,Aachen Std Bold']
['Bold,Regular']
where I can't imagine the two being synonyms. But
$ fc-list : family style | grep "Aachen Std"
Aachen Std,Aachen Std Bold:style=Bold,Regular
seems to confirm LilyPond's report.
I have the feeling I'm running against the same walls I had been
running when I tried to enable LilyPond to load notation fonts
from
system installed fonts (which failed because fontconfig refused
to
admit that when it doesn't find an exact match and instead
insists on
returning a fallback font).
I believe what you are seeing here is that there is only a bold
font. Based on a web search, Aachen Std comes in a Medium and
Bold variant. As there isn't a standalone "regular" version, it
could make sense for the bold and regular styles to map to the
same font.
I *think* I've finally understood how to read the output:
- The "names" part of the details has always one or two entries,
never more.
- The first of these names is always identical to the family
name
- If a second name is present this is the font weight this entry
is about
['Helvetica LT Std', 'style=Bold']
['Helvetica LT Std,Helvetica LT Std Cond Blk', 'style=Black
Condensed,Regular']
['Helvetica LT Std,Helvetica LT Std Black', 'style=Black
Oblique,Italic']
Each line refers the the 'Helvetica LT Std' *family*
The first line refers to the default font weight, the second and
third ones to alternative ones. In these the first element is only
there to match the family while the second is for the concrete
current font. So they are not "synonymous" as I previously
thought.
The third line refers to the "Black" weight and the "Black
Oblique" face, with "Italic" being an alias.
This also exlains the example of the Aachen Std font where we
don't need to assume that Aachen Std and Aachen Std Bold would be
synonyms. And it doesn't contradict the other cases where a font
face could be addressed by keywords in many languages.
I'll test if I can get the stuff to parse into a clean dictionary
...
Urs
I
agree that font matching is confusing at times. Sometimes you can
get away with partial names and get the right match, other times
you get an unexpected fallback. fc-match can help when testing
this, but otherwise it seems like you are best including as
complete a font name as possible.
In fact, if you had installed the other EB Garamond fonts, then
there could be confusion between EB Garamond 08 and EB Garamond
12. Based on the original list you provided, omitting the number
would seem to be allowed and it would use 08. But I'm curious
what output you'd get if you had both 08 and 12 installed.
-- Aaron Hill
_______________________________________________
lilypond-user mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/lilypond-user