Message-Id: <9311080734.AA05782@sc.ZIB-Berlin.DE>
Reply-To: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
In-Reply-To:  Message of Sun,
              7 Nov 93 21:58:00 -0700 from <michal@GORTEL.PHYS.UALBERTA.CA>
Date:         Mon, 8 Nov 93 08:33:15 SET
From: Goossens Michel <GOOSSENS%CERNVM.bitnet@vm.urz.Uni-Heidelberg.de>
Sender: Mailing list for the LaTeX3 project <LATEX-L@vm.urz.Uni-Heidelberg.de>
To: Multiple Recipients of <LATEX-L@vm.urz.Uni-Heidelberg.de>
Subject:      Re: the name NFSS
Status: R


>>
>> i'm with Frank. leave NFSS as it is.
>>
>> sebastian rahtz
>>
>
> Just to avoid misunderstandings.  I have nothing against NFSS name.
> I was protesting against prefixing with 'nf' {\em style} names for
> reasons I made abundatly clear, I hope.  So far nobody came strongly
> in a support of the contrary position, although the discussion started
> to diverge in many more or less tangential directions. Usual,
> I presume. :-)
>
>    Michal Jaegermann

As Michal seems  to say that there were no  voices _against_ prefixing
the style  names, I  shall speak up,  since, working  closely together
with  the LaTeX2e  development team,  it was  me (amongst  others) who
_suggested_ the  change.  The  reasons are  mainly practical,  i.e.  I
have  to  coordinate  (at  CERN)  the maintenance  of  LaTeX  on  many
different platforms, and  changes introduced on one system  have to be
done on  all others in  synch, otherwise  users will find  a different
behaviour for LaTeXing their documents.  As always, thanks to Big Blue
IBM, I  am limited by  the lowest common  denominator, in my  case IBM
VM/CMS with its nice 8 character limit on filenames.  So, I _am_ aware
of that limitation on some  systems (notably also ubiquitous DOS :+{).

Moreover,  VM/CMS  being  _older_  than   DOS,  it  does  NOT  have  a
hierarchical file structure, thus all files  are on the same level, so
that I  cannot "hide" files with  the same name, but  corresponding to
different versions of LaTeX in different directories (anyway, I do not
want to do a sub-directory search  on UNIX either, since it slows down
the  startup time  of  TeX enormously,  and since  most  of our  users
typeset mainly small documents, that is too high a price to pay).

So I prefer  to have the name  of the font indicate to  which class of
packages it  belongs.  And,  if you look,  then Timothy's  van Zandt's
popular pstricks  and seminar  packages have file  names limited  to 8
characters and  all starting with  pst- or sem.  I  therefore consider
the prefix  nf (New Font or  whatever) as a means  to know immediately
that I am dealing  with files to do with fonts  (probably nf should be
lf, for "latex  font").  Now your argument _mainly_ is  based upon the
important issue of "backward compatibility".   One seems to think that
when one  has a  file times.sty,  then everything  is well,  since old
files can  be run  unaltered (sic).   Now, what  is wrong  with having
files with  a given prefix (avant.sty  does not give me  the slightest
clue that this is  a package that has to do  with the avantgarde font,
while   nfavant.sty,   or   lfavant.sty,  conveys   that   information
immediately).  Also  for installing the  system, it is much  easier to
copy  all files  associated  with the  package  together, rather  than
having  a  list   with  times.sty,  avant.sty,  helv.sty   (or  is  it
helvetic.sty, helve.sty?)..   Therefore, I  propose to _leave_  the nf
prefix (or the lf prefix), and  provide files for those (two or three)
cases, which are now in common  use e.g., times.sty can have something
that  either only  contains  a pointer  to nftimes,  or  else, a  more
elaborate  scheme.   This  method  also  offers  the  possibility  for
software  maintainers to  locally customize  their times.sty,  without
touching the distributed nftimes.sty.

Moreover, I would like that principle  of unique prefix to be extended
to other  packages too, like  all the  babel option files  could start
with  lb  (or bl).   The  same  for  the  different option  files  for
LaTeX2e's  graphics, who  could have  lg (latex  graphics), etc.   The
names of these files are of no concern to the average user.

Thus my proposal  would be to _leave_  a prefix (or replace  it with a
suffix,   e.g.,  timeslf.sty   if  you   prefer),  and   have  a   few
"compatibility"  files,  like  times.sty,   which  can  be  customized
locally, and which by default  would contain merely an input statement
for  the   relevant  NF  (LF)   file.   This  scheme   offers  optimum
possibilities  for future  extensions,  eases  maintenance issues  and
provides an  adaptable way for guaranteeing  backward compatibility by
adding "bridging" files, as they are needed.

I  hope  that  (almost)  everybody  can  agree  with  this  compromise
solution.

Michel
-----------------------------------------------------------------------
 ,=====,======  Dr. Michel Goossens      Phone: (+41 22) 767-3363
/  ,----\, |    Documentation Section    Fax  : (+41 22) 767-7155
|  C E R N |\|  AS Group - CN Division   Email: Michel.Goossens@cern.ch
|\|      / \ |  CERN European Laboratory
| "====="  /\        for Particle Physics
    '--/--'  \  CH-1211 Geneva 23
      /       \ Switzerland
-----------------------------------------------------------------------