pro/contra PS resource databses

[Piet Tutelaers <P.T.H.Tutelaers@urc.tue.nl>]

Here is a summary to the questions regarding the use of the PS
resource database in ps2pk15:

1) Berthold:
 > ptmr8r Times-Roman "TeXBase1Encoding ReEncodeFont" <8r.enc putr8r
 > Utopia-Regular "TeXBase1Encoding ReEncodeFont" <8r.enc <putr8a.pfb

   or

   tir Times-Roman "TeXnANSIEncoding ReEncodeFont" <texnansi.enc <tir.pfb
   uvc UniverseCondensed "TeXnANSIEncoding ReEncodeFont" <texnansi.enc <uvc.pfb

   :=)  Please make it flexibly enough to work *not* just for 8r...

No problem. Filenames are lookuped via their internal PostScript
name. In your case via TeXnANSIEncoding (instead of TeXBase1Encoding). 
The PS resource database should contain the info to locate the files 
for example:

   Encoding
   TeXBase1Encoding=/usr/local/tex/dvips/8r.enc
   TeXnANSIEncoding=/usr/local/tex/dvips/texnansi.enc
   .

2) Pierre
To put it more generally, when you use PSRES for both xdvi and dvips,
is it possible to let xdvi (through mtpk) know about the pfa file but
still not force dvips to download it even when it is not needed?

   The mechanism Adobe uses to find out if a PostScript font is resident
   on a printer is not via the PS resource database but via the proper
   PPD file. In case of our HP LaserJet 4Si we have
      *DefaultFont: Courier
      *Font AvantGarde-Book: Standard "(001.006)" Standard ROM
      *Font AvantGarde-BookOblique: Standard "(001.006)" Standard ROM
      *Font AvantGarde-Demi: Standard "(001.007)" Standard ROM
      ..... 

   So if you have the ability to add fonts to a SCSI disk you should update
   the PPD file to reflect this situation. Of course dvips needs to
   read this PPD file to find out what font is resident instead of
   of using its current `<filename' approach while xdvi (maybe using
   mtpk and ps2pk) always needs to get that font resident or not.

3) Karl
   in favour of:
      ptmr8r Times-Roman "TeXBase1Encoding ReEncodeFont"
      putr8r Utopia-Regular "TeXBase1Encoding ReEncodeFont"

Sorry, but it's not at all obvious to me that this is an improvement.
I know you have a different opinion, but ...

What *would* be cleaner, IMHO, is a syntax that specifies the files and
their purpose, for example:

putr8r Utopia-Regular (encoding "8r.enc") (type1 "putr8a.pfb")

The software would be perfectly capable of reading 8r.enc and deducing
the TeXBase1Encoding.

   The syntax of psfonts.map
      "use of resource1" "use of resource2" <filename1 <filename2
   or
      "use of resource2" "use of resource1" <filename1 <filename2

   is perfectly suited for dvips itself. But not for other applications 
   like mtpk for example. Your suggestion would be an improvement for the 
   last category. But given the fact that Adobe's PS resource databases 
   provides a standard documented and free available way to lookup the 
   filenames of resources why shouldn't we use them in dvips and other
   PS utilities?

   The benefits are best understood if you imagine the work somebody has
   to do in order to use his ATM fonts with dvips. Should he rename all
   his fonts to the standard CTAN names? Or make a copy of his fonts
   just for working with TeX? With PS resource database it would be
   just a question of running mkpsres.
   
One thing that is not just syntactic sugar is some way to specify
do-not-partially-download-this-type1. I was going to invent something
like <<minion.pfa or <!minion.pfa or some such.

   If there is a good reason to partially download one font but not
   another one I would think of finding an extension to the PPD format
   to cope with it. Then other applications could eventually use this
   feature too.

--Piet