[Fontconfig] Aliases

[Ambrose Li]

Sorry for the tone of my previous mail.  I hope my tone this time
is better and what I write below makes sense.

On Fri, Apr 02, 2004 at 12:52:10AM -0500, John A. Boyd Jr. wrote:
> 
> - "Times", "Timmons" - symbolic, not semantic/descriptive
> - "Serif", "Sans", "Monospace" - semantic/descriptive

How about names like "Baskerville" or "Caslon", etc. which are
in the public domain and somewhat cross the boundary between
symbolic and semantic?  I agree with you that they ultimately
are shorthands for lots of descriptive attributes; however,
I would say that it should not be entirely dismissed that
certain names can legitimately / usefully serve as abstractions
themselves.
 
> How about "non-Latin" for starters?  Then, say, "Hebrew",
> "Greek", ... (These are character set descriptions, I'm
> imagining, from my limited knowledge of fonts.)

I think fontconfig already knows something about this.

What I was trying to say was, e.g., the Kai (CJK) style being
classed by default as "sans serif" by stock fontconfig and as
"Kai" in Debian. According to limited understanding of how
Latin fonts are categorized, both classifications are wrong; it
should properly be classed as italic even though it would not
match well with Latin italic types (because of different italic
angles).  (And naturally, the panose data are wrong about the
correct classification.)  Continuing with this particular case,
what do we do with existing classification schemes, and how
far do we go in making sure something is correctly classified
using existing criteria (in this case the correct definition
of "italic"); and how do we handle unexpected cases where two
different styles *ought* to be classed identically but are
incompatible?

Or perhaps let us consider the "Kai" style from a more realistic
point of view: In a normal fontconfig installation, when the
user asks for a "sans serif" Chinese font, the system is likely
to deliver a Kai font because of the simple reason that there is
no free Chinese sans serif type in existence.  But in this case
a typical user -- provided that he/she knows what "sans serif"
means -- would be able to tell that this is wrong, if not only
for the reason that Kai types have (round, sometimes subtle)
serifs.

> Let's try another example.  Let's say we want to describe
> font width - I like that example...  How about "condensed,
> expanded," and the like.  I'll bet you can find these in
> existing fonts as well.

Yes. But let us pause for a moment and say the same thing
about "normal", "medium", and "regular".  These certainly sound
identical, mean the same thing to most people, and might even
mean the same thing for lots of fonts.  But there are also lots
of commercial fonts where they represent different weights.

And there are commercial fonts who describe themselves as
"normal" or "regular" (say) even though they are bold (say).
We should be aware of such irregularities.

> I would suggest, Ambrose, that you're thinking less like a
> user than like a system developer.
> 
> In that regsrd, let me just ask: how does (or might) a _user_
> discriminate fonts at the level of panose information?

I'd say there are different kinds of users.

There are users who don't know what sans serif is (or what
a serif is), the same users literally would call all any old
sans serif font "Arial" (and not know what "Helvetica" is) and
all serif fonts "Times New Roman"; and I personally do know a
number of such users.  To these users, even serif and sans serif
are meaningless descriptions.

At the other extreme, there are users who know what "FF Meta"
and "Myriad MM" are, and some who can tell the difference
between Helvetica and Arial.  These users (those who have
studied typography and/or do typographic work day to day) would
at least have a handle on what the panose data mean.

Of course, the two kinds of users would discriminate fonts at
the level of panose information differently.  The "clueless
users" might be able to only distinguish between panose "family"
and weight, the "professional users" will be able to distinguish
most of the panose attributes, and the "academic" might be able
to distinguish all.  The "professionals" and the "academics"
certainly would discriminate fonts based on more than what the
panose data would describe (e.g., by the "feel" of the font, the
"historical period", etc.), especially for some of the panose
attributes which are not helpfully descriptive (e.g., when
Family is Script or Decorative); however, their ability to
distinguish might be subconcious.

Under normal use, I might speculate that perhaps a quarter to a
half of the panose data are useful, and perhaps several
additional, non-panose descriptions will be required to
sufficiently describe a font.  Certainly, to the user, the
panose data are not sufficient to meaningfully describe fonts.

I would assume that the typical user you have in mind is
somewhere between the two extremes.  However, at least one
extreme cases is not negligible; "clueless users" do exist, and
GNU/Linux becomes more and more generally usable as a desktop
system it could attract more and more "professional users". We
can reasonably ignore the "clueless" ones, but what I call the
"professional" users might describe type in a way you would call
"like a system developer".

If what we are going to do is to improve on the way fontconfig
is to do matching, by using more descriptive attributes, I have
no objections; this can only benefit typical and professional
users alike.

> Aliases have the properties of abstractions if they are also
> abstractions, but again, they are then only a special case.
> But in general, aliases are not also abstractions.  In general,
> an abstraction can be used wherever an alias can, but the
> opposite is not true at all.

I would tend to be a bit reserved about this statement. Unicode
(at least the CJK unification part) is supposed to be based on
the same principle, but in practice it does not deliver even
though this principle should be theoretically sound.

I might speculate that this property would break down for script
and decorative types, esp. the latter.

Descriptive attributes may change in time too.  Helvetica at one
time was perceived as "authoritative"; today perhaps "default"
or "plain" might describe it more appropriately.

> To your second point - no; there's a legal principle called
> the doctrine of fair use that applies to certain classes of
> use of intellectual property.  Teaching, journalism, and
> research are among those uses exempted under the fair use
> doctrine, as I understand it.

I am not as optimistic.  Judging from how the recording, movie,
and commercial software industries, and our legal system are
acting, I would say that fair use will disappear in the future,
perhaps the near future, or perhaps it already disappeared
behind our backs.

I certainly hope that I am just being overly pessimistic.