Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!uunet!snorkelwacker!apple!apple.com!kevina
From: kevina@apple.com (This space for rent)
Newsgroups: comp.lang.postscript
Subject: Re: Pathforall returns "undefinedresult"
Message-ID: <9986@goofy.Apple.COM>
Date: 30 Aug 90 15:32:06 GMT
References: <1990Aug30.004154.23677@metro.ucc.su.OZ.AU>
Sender: usenet@Apple.COM
Organization: Apple Computer, Inc.
Lines: 29

In article <1990Aug30.004154.23677@metro.ucc.su.OZ.AU> 
rossc@extro.ucc.su.oz.au (Ross Cartlidge) writes:
> ...
> Part of the "conversion" is mapping paths and clippaths - this uses
> pathforall. Someone has given me a PostScript file which, when
> multi.ps is prepended, gives an "undefinedresult" for operator 
"pathforall"
> 
> What conditions can make this happen?

PostScript paths are mapped to device space (i.e. the resolution of the 
underlying pixels) as they are created.  However, operators like 
"currentpoint" and "pathforall" operate in user space.  In the case of 
pathforall, each point in the path must be mapped from device space to 
user space (as if by itransform) and placed on the stack for use by the 
appropriate callback procedure.  

A "undefinedresult" error probably means that somewhere in the combination 
of the original program and multi.ps, a singular (non-invertible) current 
transformation matrix (CTM) is getting established.  The net result is 
usually a divide by zero error.  (Either that or your points are getting 
out of range, but most Adobe interpreters are fairly lenient about 
infinity conditions.)  One way to debug is to print out the matrix before 
pathforall is called and see if "invertmatrix" works on it.

Good luck!

--Kevin Andresen [kevina@apple.com]
"It's a small world, but I wouldn't want to paint it."