John A. Boyd Jr.
jaboydjr at netwalk.com
Thu Apr 1 03:48:02 EST 2004
First, I agree with Keith that trademark names are an issue, but
it must be considered that there are both registered and unregistered
trademarks. It would be nearly impossible to protect against using
In any event, if words like "Monospace" are trademarks, the last thing
one would ever want to do is hardcode them into software; that makes
the author of the software the infringer, not the user. If the choice
to "infringe or not to infringe" is to be left to the user, the use
of such names must be derived from the processed fonts, not hardcoded
in any way.
But to Keith's proposal, may I suggest an alternative to the use of
"aliases". I think semantically, these are not "names", and thus not
"aliases" at all; they are generic descriptions of style. Following
from the motivations for the patches I submitted some time ago, these
could be considered "anonymous" patterns (in the fontconfig sense),
where no attempt is made to match on name, but matches are instead
done on style (as rendered by semantically distinguishable keywords).
E.g. what is now referred to as the "Sans alias" could instead be
considered the "sans" style keyword for any matching font name
(i.e., family). Moreover, it can be derived in many cases from the
underlying font description without having to hardcode it.
I didn't go quite this far with my patches, though I did begin to
establish the framework for dealing with style as a collection of
keywords (not just in fontconfig, but in pango, freetype, etc.).
I offered them as a workaround, but one with a clear direction towards
Notably, this direction avoids the "author as infringer" dilemma,
since it avoids hardcoding of keywords which might also be trademarks.
Note that the use of style keywords need not be imagined to, e.g.,
simple case-insensitive string matching. More general mappings should
be possible as well. I.e., whatever is now done to map aliases could
be done as well to map style keywords, since after all, the current
aliases _are_ style keywords, though less generally interpreted than
is possible and semantically reasonable.
Keith Packard wrote:
> Around 0 o'clock on Mar 31, Juliusz Chroboczek wrote:
>>I would tend to agree with Adam here. Preventing the user from
>>infringing on trademarks is not fontconfig's business; it's the
>>business of law enforcing agencies.
> Of course; what we want is to allow people to get their work done without
> forcing them to infringe on trademarks. Fontconfig does half of that job
> already; providing a way to accept trademarked names and match them
> apprproiately while allowing people to leave those same names out of the
> application font menus.
> Now it seems like we've figured out that we want the other piece as well;
> providing a way to get new names to appear in the font list which are then
> mapped to other fonts. I can imagine this would be really nice for
> presenting localized font names, as well as the existing generic stuff
> that gnome has kludged into place.
> Let's try to get some use models and proposed semantics set out so we can
> start figuring out how this should work.
> fontconfig mailing list
> fontconfig at freedesktop.org
More information about the Fontconfig