Re: a DSSSL typesetter

[Brandon Ibach <bibach@xxxxxxxxxxxxxx>]

Frank A. Christoph said:
> 
> By the usual Turing completeness argument, it's not hard to see that
> anything you can do in TeX, you can do in PostScript.  The real issue is,
> _where_ you do it: on the computer, or at the printer?  As anyone who was
> had experience producing PostScript code before knows, the trick is to do as
> much computation as possible on the host machine, and then send the
> smallest, least computationally-intensive result possible to the PS file or
> printer.  (Rather like partial evaluation in a functional language---the PS
> file is the residual program.)  The reason is not only that PS interpreters
> are slow and low on memory, but also the fact that you can obtain greater
> guarantees about the resulting document if you do as much computation as
> possible as early as possible.  For example, TeX tells you when you have
> dangling references (to labels), it tells you when it can't find an external
> input file, it tells you when there is an error in the macros, it tells you
> when a line is overfull, etc.  So you can take corrective action before you
> actually send the thing to a printer.
> 
   A good point.  As I said, though, I could see the typesetting
process happening at the computer, instead.  Basically, the
typesetting code would emit simple PostScript code that could easily
be handled by a printer without straining processor or memory
resources.  In effect, the PostScript typesetting routines would write
a new, simple PostScript program.  And, of course, any errors
encountered during this process could be directed to the user.

-Brandon :)


 DSSSList info and archive:  http://www.mulberrytech.com/dsssl/dssslist