Re: TrueType & Fontname

[Lars Hellström <Lars.Hellstrom@math.umu.se>]

It is clear (and has been for some time) that the current Fontname scheme
is insufficient---Werner Lemberg's comments about TeX names for CJK
subfonts is yet another sign of this---and the main problem is still the
requirement that the TeX font names must be valid file names in all major
file systems. I say let go of that! If names up to 20--24 characters are
allowed, solving all these naming problems becomes simple for the
foreseeable future (and most likely a good deal longer).

And yes, we've had this discussion before. It is quite true that not
everyone is on a file system that can handle names that long and neither
can be expected to buy one, but there are systems (e.g. kpathsea in web2c)
which can perform non-identity mapping of TeX font names to file names. The
TeX Live CDs now contains long filenames (in at least two different
formats) for files not fonts, so why should it not be possible for fonts as
well?

The mere existence of a scheme for long font names does not make the short
names unnecessary. In the case a mapping of long font names to shorter file
names is employed, one still has to have a scheme for short names, even
though they might be entirely site-specific. Using the Fontname scheme as
far as possible and assigning arbitrary names consisting entirely of digits
to resolve any collisions or insufficiencies in Fontname that are
encountered should work fine.

Furthermore, if a system for long names is designed so that no long name is
shorter than nine characters, then nothing says one cannot use the current
naming scheme for some names and the new scheme for others. In particular,
those who are on old systems AND cannot use kpathsea (or something similar
for name mapping) AND produce files that use non-CM fonts AND send these
files to other people can STILL continue to use the current Fontname scheme
without any risk of fonts being confused. That's more than one can say
about the current situation concerning Fontname names and names of various
MF fonts.

The design of a scheme for long font names should be begun, the sooner the
better.

As for Walter Lemberg's suggestion, I can see nothing wrong with it, but
one simply cannot continue to add patches (after all, that's what it would
be) to the current Fontname scheme for much longer. The partitioning of the
namespace is now alarmingly far advanced.

Lars Hellström