Comments wanted: Directory structure of fontinst/inputs/

[Lars Hellström]

Hi, all.

As no new bugs have been found in fontinst v1.914 for about two months, I
believe it is safe to make v1.915 (which is little more than a corrected
v1.914) the non-prerelease version of fontinst v1.9. Hopefully it shouldn't
have to take that long, but of course I can't make any promises ...

One thing that I think should better be discussed first is however the
directory structure of fontinst/inputs/, as the current structure is far
from helpful---it provides no more information than what you find in the
file suffixes! I believe it would be nice to organise the files so that you
get some idea about what kind of font it is indended for, as it may
otherwise be hard to find the file that (hopefully) does what you want, so
what follows below is a suggestion for such a directory structure. To
illustrate what would probably go where, I have inserted the names of all
files that appear in fontinst/inputs/ or fontinst-prerelease/inputs/ in
what would be their places according to this suggestion.

base/
   cfntinst.sty
   finstmsc.sty
   fontdoc.sty
   fontinst.ini
   fontinst.sty
   trig.sty
   xfntinst.sty
latinetx/
   8r.etx
   8y.etx
   ot1.etx
   ot1c.etx
   ot1cj.etx
   ot1ctt.etx
   ot1i.etx
   ot1ij.etx
   ot1itt.etx
   ot1j.etx
   ot1tt.etx
   t1.etx
   t1c.etx
   t1cj.etx
   t1i.etx
   t1ij.etx
   t1j.etx
latinmtx/
   8r.mtx
   8y.mtx
   latin.mtx
   latinsc.mtx
mathetx/
   euex.etx
   eufrak.etx
   eurm.etx
   euscr.etx
   oml.etx
   oms.etx
   omx.etx
   rsfs.etx
mathmtx/
   mathex.mtx
   mathit.mtx
   mathsy.mtx
smbletx/        (SyMBoL)
   ts1.etx
   ts1i.etx
   ts1ij.etx
   ts1j.etx
smblmtx/
   textcomp.mtx
misc/
   csc2x.tex
   csckrn2x.tex
   glyphoff.mtx
   glyphon.mtx
   kernoff.mtx
   kernon.mtx
   resetosf.mtx (?, perhaps smblmtx)
   resetsc.mtx  (?, perhaps latinmtx)
   unsetalf.mtx (?, perhaps latinmtx)
   unsetnum.mtx (?, perhaps smblmtx)

As you can see, I have maintained the ETX/MTX distinction but additionally
distinguished between latin, math, and symbol fonts. (Thus the current
West-European preponderance in terms of fontinst support would be more
clearly displayed as just a property of the support, and not of the program
as such. BTW, didn't Vladimir Volovich some time ago write something about
that he was writing support for cyrillic fonts? IMHO that seems like a good
starting point for two directories cyretx/ and cyrmtx/!)

The base/ and misc/ are there for a slightly different reason: I thought it
might be a good idea to organize the contents of fontinst/inputs/ so that
they can simply be copied into the tex/fontinst/ of a TDS tree. If I recall
things right (which is not at all certain, as my experience with TDS
systems is quite limited) then the TDS specification says that these two
directories must be present, so they might as well be used. Keeping the
current tex directory probably wouldn't be a good idea.

Another thought in this is that a more diversified directory structure
might make it easier to distribute maintenance responsibility (with which
would of course follow the corresponding authority). Even though I have
been able to find quite a lot of time for working on the actual fontinst
program (i.e., pretty much what appears in base/ above) during the last two
years, I haven't found much time for maintaining the rest of the system,
and neither have any of the other authors of fontinst. It follows that we
should probably allow other people to assume responibility for the rest of
the system, and I suspect that it is a more reasonable task to take over
latinmtx/ than to take over mtx/ (especially if you have little experience
with, say, math fonts).

But this is only a _suggestion_. I want to know if it is any good.

Lars Hellström