Postscript Parser

[Elizabeth Zwicky]

In article <10035 at brazos.Rice.edu> loki at physics.mcgill.ca (Loki Jorgenson Rm421) writes:

>Is there a "lint"-like parser for PostScript which will check for and
>identify syntax errors in a PS file?  Preferably Public Domain, of course.

This is one of the two classes of questions that people ask frequently on
comp.lang.postscript that indicate a fundamental misconception about
PostScript. (The other class has as its worst example "Does anyone have a
PostScript to ASCII translator" and indicates a much *worse* fundamental
misconception.)

The problem here is that almost all of the interesting part of PostScript
is semantics, not syntax. The only syntax rule states that unquoted
parentheses may not occur inside strings. (It is legal, if unadvisable, to
end a file without closing open delimiters.) The number of true syntax
errors people make is vanishingly small, and they're all easy to catch.
Even throwing in a check for delimiter-matching doesn't help much, since
delimiters can legally and logically be crossed (i.e. it is possible to
have a reasonable program which mixes delimiters like this: { (}) [} {(])
] (}) }, without quoting them, which is going to drive anything that tries
to match delimiters via pure syntactic rules straight up a wall.)

In short, lint for PostScript is a) trivial and b) worthless. What you
need instead is an interactive debugging environment, to let you get a
handle on the semantics. If you're programming a printer, this is a real
problem; if you're programming a display, you should be able to find some
such tool. (Actually, I tend to debug printer code on a display, for this
reason.)

	ELizabeth