diff options
author | Philipp Gesang <phg42.2a@gmail.com> | 2013-06-22 16:16:21 +0200 |
---|---|---|
committer | Philipp Gesang <phg42.2a@gmail.com> | 2013-06-22 16:16:21 +0200 |
commit | bd0a8039e89b9403b58ac5a9827a333276a6fbf6 (patch) | |
tree | 8e8cc079a4b9b235df95344a30873de969f497fa /src/fontloader/luaotfload-fontloader.lua | |
parent | 5fa488d55de0b89708e6028491fffa5f43ca279d (diff) | |
download | luaotfload-bd0a8039e89b9403b58ac5a9827a333276a6fbf6.tar.gz |
treat *medium* as a synonym for *bold*
treating *medium* as *regular* weight leads to incorrect matches with
Garamond Premier and likely other fonts as well. apparently, this also
happened with other typographical programs, e. g.
http://blogs.adobe.com/typblography/2008/05/indesign_font_conflicts.html
note: the matching algorithm could be rewritten so as to resolve names
in two passes:
1. collect a list of candidates. if there is a literal match, pick that one
2. test style synonyms against the candidates according to a precedence list so that “regular” always precedes “medium”
to take this thought a bit further, the synonyms could also be ordered
by weight according to some assumptions. then we could pick the closest
weight present in the database. example: suppose *regular* is assigned a
weight index of 3, *medium* gets *4* (i. e. just a tad heavier), and
*bold* 7. then the synonym regular would be preferred for the request
*regular* if present, but medium would be chosen over bold if there is
no *regular* shape. this would probably be overkill, though.
Diffstat (limited to 'src/fontloader/luaotfload-fontloader.lua')
0 files changed, 0 insertions, 0 deletions