Path: utzoo!utgpu!jarvis.csri.toronto.edu!clyde.concordia.ca!uunet!intercon!heaven!glenn@NeXT.COM
From: heaven!glenn@NeXT.COM (Glenn Reid)
Newsgroups: comp.lang.postscript
Subject: Late night thoughts after reading comp.lang.postscript
Message-ID: <1692@intercon.com>
Date: 8 Jan 90 17:21:39 GMT
Sender: amanda@intercon.com
Reply-To: heaven!glenn@NeXT.COM (Glenn Reid)
Organization: NeXT Inc.
Lines: 94

[Note: I am posting this for Glenn from my machine, since he is evidently
 having problems posting directly.  Feel free to verify with him that it's
 not a forgery :-) --Amanda Walker]
---------------------------------------------------------------
This is a response to a whole slew of things I've seen on
comp.lang.postscript in the past month or two. Forgive me for
bringing up a whole bunch of unrelated things at once, and
for jumping into the fray, but I am disappointed at the level
of noise that has been clogging the arteries on what used to
be a wonderful newsgroup.

First order of business:  Woody, I'm sure you are a wonderful
hacker and very proud of your accomplishments, but almost
everything you post to the net has nothing to do with
"comp.lang.postscript"  You are consistently well outside the
boundaries of the PostScript language, as Amanda and others
have been kindly and gently pointing out.  Maybe it would be
best for you to post your 20 messages per day to comp.laser-printers
or wherever it is that they discuss printer internals and
undocumented hacks and downloading machine code and stuff
like that.  I'm sure it's useful stuff, but it's useful to
one person, in one location, for one printer, for one purpose.
And if I were a "customer", I would be very, very worried about
what happens when you aren't around to support the code and
things quit working, since none of it is written down anywhere
and much of seems dependent upon the dark underbelly of computing.

Also, you need to understand what an "undocumented feature" is
before you get your programmer's stripes.  No one is
being sneaky and deliberately hiding useful stuff from you.
An undocumented feature is something that happens to work but
is unsupported and may not work in the future or anywhere else.
These are not documented precisely so that you won't think that
they are part of the language and start using them.  They are
left undocumented for your own protection.  If you can figure
out how to use them, great.  But don't flame the company for
not documenting them, because they shouldn't be.  Every computer
company on the face of the earth has to make that decision, and
Adobe has done an excellent job of drawing the line between the
language and the printer implementation.

You know, I have worked with PostScript intimately for over
five years, and I know virtually nothing about secretdict,
$printerdict, mumbledict, ROMnames, or any of the other
undocumented features of PostScript.  And I'm proud of that.
There is no benefit that I can think of to knowing.  It's like
programming the Macintosh entirely without using the Toolbox,
so that your program is guaranteed to break when they bring out
the new system.  It is like taking advantage of strange bugs in
your C compiler that you know won't work on another compiler.
It's like trying to write a portable program in assembly
language.  There just ain't any sense in it, and whatever benefit
there is is very short term and not worth the time, energy, and
money invested in it.  A $5,000 laser printer is cheap compared to
a good programmer's time, and I'm awfully tired of reading about
it.  I suppose that in fairness, maybe other people are getting
something out of the conjecture, opinion, and misspellings that
I have missed and that I shouldn't complain, but I will anyway.

Now, on to other things.  Someone asked why PostScript was
designed the way it was, and somebody else asked how one might
be able to print a 1 megabyte file or something very big.  Well,
they have the same answer, really.  PostScript is interpreted
and stack-based so that it can be executed on the fly directly
from the input stream, without having to buffer lots of stuff
or maintain much state in the interpreter.  The program doesn't
have to all fit into memory at once.  You can execute arbitrarily
large programs that way, and that's why it was done.
If a program is reasonably well written, you should be able to
make marks on the page all day long without exhausting the
memory or resources.  The early versions of Adobe's Colophon
newsletter often were 6-10 megabyte PostScript files with images
spanning a 17x22 inch spread, lots of downloaded fonts, etc.
A little prudent use of save/restore and making sure to print
images according to the example in the Red Book is all it takes.

I think there were going to be more content to this message, but
now I've managed to forget whatever it was that I was going to
say.  Oh, here are a few of them:  "filenameforall" is supposed
to supply a file name each time around the loop, otherwise it
is useless; Adobe doesn't make anywhere near $1000 per printer
(also, you don't double the royalty because the purchase price is
double the manufacturing cost, the royalty is part of the
manufacturing cost); 


In any case, let's bury the whole issue of undocumented code
and talk about something more interesting, or else let's go
find a net group that has no traffic and talk about PostScript
there, instead.

Glenn Reid
    speaking only for myself, from home, trying to rescue and
    preserve the spirit of the fine PostScript language