[Fontconfig] Aliases

[Enrique Perez-Terron]

On Thu, 2004-04-01 at 10:13, Keith Packard wrote:
> Around 3 o'clock on Apr 1, Enrique Perez-Terron wrote:
> 
> > As far as I understood this issue, it started out with a request to be
> > able to list the font names that an application can "ask for" without
> > being turned down.
> 
> No.  That capability is already very well supported.  That let's people 
> say "if an application asks for 'Times', please use 'Timmons'", it does 
> this in a way which doesn't expose 'Times' in the list of fonts that 
> applications place in menus, which allows people to avoid trademark 
> problems.

I realize that a user that composes a new document and looks in a menu
for available fonts will expect that the listed fonts are available, not
just approximable. This means that even the menu itself does not contain
complete sentences with subject and predicate, the application might be
seen as making a false, and indeed infringing, statement, or using the
name in an infringing way.

So far, I was probably too quick to dismiss the problem.

On the other hand, I still see it as reasonable to ask "Which fonts does
the system have sufficient knowledge about to pick a reasonable
substitution", or "For what font names has the system administrator
configured a response".

What if fontconfig gets an added interface that is documented as
answering these questions? The documentation could even urge the
developers to consider that some ways of using this information could
lead to trademark infringement issues.

If then the application developer misuses this interface to give the
users of his application false or infringing impressions, would that be
his responsibility, or could even fontconfig's developers be held
responsible?

If fontconfig distributes a list of fonts that are believed to be
reasonable substitutions for one another, is that too an infringement?

I am quite convinced that I can safely say in public that "while a
friend of mine used a brand X cleaner I found that ordinary soap is just
as effective", without this statement constituting an infringement of
trademark X. So, if it is possible to quote protected names in
non-infringing ways, where exactly is the limit? I still feel confident
that expressing "Font Y(tm) can be substituted for font X(tm) with
reasonable results" does not by itself constitute an infringement.

If I prepare a document, and include in it codes that indicates to
viewer applications that fonts X, Y, Z should be preferred for
displaying it, if available on the reader's computer, do I need to have
specific licenses or rights w.r.t the named fonts? I believe there are
many non-infringing uses of trademark names.

Anyone with more specific knowledge?

> If an application says "I have the Times font available" when it really
> means "I have a font that looks just like Times available", that represents
> a clear trademark infrigement.

Yes.

> So, it's nice that we have this ability to do sensible font substitution 
> without forcing people to violate trademarks like the XLFD mechanism does.
> 
> However, it would *also* be nice to allow people to add names to 
> application font menus and have those names directed at sensible 
> substitutions.  Then we could add things like 'Sans-Serif' and have it 
> displayed in application menus without having special application kludges 
> everywhere.  Of course, this capability would let people add 'Times' to 
> their application menus without really having the 'Times' font, but that's 
> not something we should even try to prevent.

There is a question what is the statement really, when there is no
statement structure, just a name in a list. The statement is then
implied, but how do we know what it is? I guess this would be determined
from circumstances and actual effects, e.g., impressions effected in the
users. ("Is Times really so ugly"?) Other possible interpretations of
the list could be:

    - This program can prepare documents that are reasonably
      rendered using font X (if the reader has it)
    - This program can show documents prepared for use with
      font X

> This mechanism could be as easy as a list of family names which would get 
> returned whenever an application requested the available families; 
> figuring out how that semantic would work will be a bit tricky as the 
> listing semantics really only deals with "real" fonts that have files and 
> character sets and the like.
> 
> -keith

Regards, Enrique