From: Kevin Brown (graphity@adelaide.on.net)
Date: Tue Dec 31 2002 - 05:35:41 EST
As if font developers haven't got enough new stuff to deal with, now
Adobe have just updated the Adobe Glyph List. (v2.0 September 2002). It
is now nearly FIVE times the size (4,910 names) compared to the so-called
"last ever" version, v1.2 October 1998 (1,050 names).
The big question is why have they done it at all? The document "Unicode
and Glyph Names"
(http://partners.adobe.com/asn/developer/type/unicodegn.html) states -
"The purpose of the Adobe Glyph Naming convention is to support the
computation of a Unicode character string from a sequence of glyphs. This
is achieved by specifying a mapping from glyph names to character strings"
OK, but what I can't find in the document is a clear statement regarding
what exactly an AGL glyph name achieves that a generic uniXXXX name
doesn't - apart from the dubious benefits of human readability. And if
there is no user advantage, then why expand the list? The new
specification also states "Font producers are encouraged to follow this
specification in naming their glyphs" Why? And what happens if I don't?
They don't say.
Was there any industry pressure for an extension to the AGL? I've
certainly never heard of any. Quite the contrary - in fact I probably
wouldn't be quite so uptight about this if there hadn't been such
apparently universal agreement amongst those "in the know" that the 1998
AGL was NEVER going to be updated. So now that a new version has popped
up "out of the blue" after 4 years I think there's some pretty damn good
reasons to resist its adoption.
My concern is that the Adobe Glyph List may well exist for the sole
benefit of Adobe, especially if they insist on making the functionality
of their applications eg Acrobat and InDesign and also their PostScript
font format partially rely on it. Perhaps Adobe are having a just little
bit of trouble letting go? Maybe they need to eek just a little extra
life out of the "old" PostScript format? But maybe if Adobe's software
can't work without alphabetic glyph names, well maybe they might just
have to think about changing their software. Shock! Horror!
Unicode names (uniXXXX) are simple, short and unambiguous. They map
one-to-one in BOTH directions with codepoints and, gee, maybe in a few
years we won't even need in-built glyph names at all (ie we'll just use
the codepoints). After all, developers of large international fonts have
got quite accustomed to most of the glyphs in their fonts not having
humanreadable names.
I know Unicode names are not particularly user-friendly but with 65,536
possible BMP codepoints I think we've all accepted by now that there are
never going to be 65,536 shorthand humanreadable names, so why does Adobe
keep creating them? And anyway, it's only ever going to be people who
speak English who will thank you for the new names (and of course since
the major part of the BMP and the world's population is not even Latin,
let alone English, the majority are not going to be thanking anyone at
all!)
When I definitely needed a human readable name for a glyph (or
conversely, a unicode value for a particular name) I simply refer to the
Unicode Names List downloaded from
http://www.unicode.org/Public/UNIDATA/. It's always instantly available
to me (via my Apple Menu, if you must know). The clear advantages of the
Unicode Names List over the AGL are (1) it's got five times the names of
the new AGL (2) the glyph descriptions are more exact (particularly as
they are not limited to 32 characters), (3) the Unicode list gives
English descriptions of the large number NON-human readable AGL names ie
those in the form "afii61573" and (4) the Unicode Names List is created
and moderated by the same body which assigns the codepoints. Given this,
I find it amazing (and unfortunate) that the AGL has achieved the
penetration it has!
There is no reason why a special version of the Unicode Names List
database (currently more than 23,000 characters - over a third of the
BMP) could not be linked to applications like InDesign and FontLab etc to
give users up-to date access to the complete set of Unicode glyph
descriptions instead of an extremely small subset (ie the AGL) of the BMP
built into the font itself. And remember, the number of names available
would be five times as many as with the AGL and the descriptions would be
more human readable. What I had in mind would be like the glyph name
database built into Windows Character Map which appears to be sourced
directly from the Unicode Names List.
The case for built-in human readable glyph names is further reduced by
the fact that the major font editing application, FontLab, by default
illustrates all glyphs in its bitmap layer so characters can be easily
accessed by reference to their codepoint and/or their physical bitmap
representation.
You might think that if the new AGL 2.0 had at least ONE thing going for
it, it would be exhaustiveness. With nearly 5000 glyph names you would
think that every last latin-based glyph would be assigned a name. But no!
There are gaps, inexplicable from my point of view, even within such
major sub-ranges as Latin Extended-B. For instance there are NO glyph
names assigned in AGL 2.0 to U+01F6, 01F7, 01F8 and 01F9. And what's
more, there appears that there is no prospect that names EVER will be
assigned to these codepoints, because Adobe states clearly in the
introduction to the specification:
"Because it is anticipated that this specification will be implemented in
many pieces of software, and that revising consistently all those
implementations is unlikely, this specification is intended to be stable,
i.e. never revised. In particular, it is intended that no mappings will
ever be added to the AGL"
Go figure! (It's not that I WANT it to revised of course. I just want it
to go away!)
The dozen or so duplicates in AGL1.2 (eg hyphen, fraction, Delta etc)
have always been one of the banes of a font developer's life and not a
small argument in its own right for the use of the "alternate" uniXXXX
name format. Adobe themselves used to suggest resolving the problems of
AGL duplicates by assigning uniXXXX names to one of the duplicates. But
now Adobe have given these former duplicates unique names. For example,
U+002D and U+00AD both used to have the AGL name "hyphen", so we used to
rename the second glyph "uni00AD". But in AGL2.0, U+00AD now has its own
name "sfthyphen". I'm sure Adobe are very proud of themselves for fixing
these duplicates (it only took 5 years!) but do we now all have to change
"uni00AD" to "sfthyphen" to make sure our fonts will work properly in
Adobe applications? And if not, I ask again why did they bother?
(By the way, I hope Adobe have got a damn good reason for calling it
"sfthyphen" and not "softhyphen"!)
Incidentally, I see that duplicates have not been eliminated entirely
from the Adobe Glyph List. But now the duplicates are in the codepoints
rather than in the names! For example, both "gravecomb" (the AGL1.2 name)
and "gravecmb" (the AGL2.0 name) both get mapped to U+0300. Sigh!
The extensive public use of the Private Use area (albeit in the Corporate
Use Subarea U+F6BD - U+F8FE) by the Adobe Glyph List continues
undiminished with version 2.0. Some of the recent almost unanimous
arguments on this list against William Overington's never-ending PUA
proposals seem pretty lame when you see the AGL's very public (and
apparently unchallenged) appropriation of a not-insignificant part the
PUA by Adobe/Apple. If Adobe and Apple feel so strongly about the PUA
glyphs they've "locked up" so publicly since 1997 then why don't they
officially apply to the Unicode Consortium to have them included in the
Unicode Standard like the rest of us have to do?
I think that the continued inclusion of these PUA codepoints in the AGL
is something that the Unicode Consortium should not remain silent on,
particularly when you consider that simply by virtue of their presence in
the AGL these glyphs now appear in the default font templates for
FontLab. In other words, these AGL glyphs are not a million miles from
becoming a defacto standard through the back door. William Overington
could not, in his wildest dreams, have hoped for a comparable result!
Finally, Adobe haven't made things any easier for font developers by
reversing the sequence of the data in the lines of the AGL compared to
earlier versions. Version 1.2 has the unicode scalar value followed by
the glyph name (semi-colon delimited). In version 2.0 the glyph name
comes first, thereby making it harder for those of us who may need to
manually modify the new AGL file so it can be read by such programs as
FontLab. Thanks a lot!
One has to wonder whether any font developers and font-editing software
vendors were even consulted during the development of this new AGL
version? FontLab 4.5 was released in late December (three months AFTER
the release of AGL 2.0) but it shipped with the OLD version of the AGL so
maybe that answers my question. I only found out about the new AGL
version today, and that was completely by accident. I guess I must be in
the wrong loop.
I have many, many fonts set up with uniXXXX names for glyphs that were
not in AGL v1.2. Now AGL v2.0 has assigned alphabetic names to many of
those glyphs, so the answer I REALLY need to get from Adobe is whether I
will need to update those uniXXXX names to AGL 2.0 names in order to
guarantee FULL functionality of those fonts (TrueType, PostScript - Mac
and Windows - and both flavours of OpenType) in Adobe applications and
elsewhere????
If the answer to my question is that the new names are not mandatory, I
for one will be more than happy to give AGL 2.0 a wide berth - but my
gut feeling is that we could be stuck with it, especially if we just sit
back and accept it.
I think that in this age of fully Unicode fonts, externally-defined
alphabetic (humanreadable) glyph names built into fonts just add to a
font developer's workload for NO GOOD REASON. And arcane mappings like
the Adobe Glyph List with its duplicates, its apparently arbitrary
omissions, its stated intention to never be exhaustive, its many
typographically inaccurate names, its mixture of humanreadable and
non-humanreadable names, its name changes from one version to the next,
the absurd interval between updates, and its public assignment of names
to Private Use Area codepoints, all make the AGL probably just about the
WORST mapping that a font developer could ever imagine. My feeling is
that if it didn't have the imprimatur of a company as large as Adobe it
would have been discarded years ago as a bad joke! I really think it's
time to bury it.
Happy New Year!
Kevin Brown
*****************************************
Kevin Brown's G R A P H I T Y !
DIGITAL TYPE SPECIALIST * GRAPHIC DESIGN
147 Magill Road, Stepney S.A. AUSTRALIA 5069
Phone/Fax: +61 (0)8 8362 8664
Email: graphity@adelaide.on.net
Est.1979
*****************************************
This archive was generated by hypermail 2.1.5 : Tue Dec 31 2002 - 06:14:40 EST