🚀 go-pugleaf

Re <9208060627.AA10179@cookie-crisp>
<1992Aug4.173917.27233@cfa.harvard.edu>:

> what i don't understand is:
> 
>   - why did gs blow up on the program?  i can't find anything wrong with it.
>     for some reason pathbbox is swallowing up two stack entries.

GS is blowing up because the second time pathbbox is executed, there
is no current point.  As a result, the nocurrentpoint procedure is
executed, pushing two values on the stack instead of the 4 that
pathbbox is supposed to push.

However, it is a GS bug that there is no current point for pathbbox.
The bug is that strokepath discards a trailing moveto.  This will be
fixed in 2.5.1.

>   - why should pstack have any effect?  i thought pstack was essentially
>     a diagnostic command with no effect on anything?

This is yet a different GS bug.  It is not pstack per se that makes
the difference, but the fact that the pathbbox is no longer the last
element of the procedure it occurs in.  After GS handles an error, if
the offending token was not the last element of its procedure,
execution resumes by re-executing the token that caused the error,
rather than continuing to the next token.  As a result, the pathbbox
is executed a second time, pushing 4 more values on the stack (in
addition to the two pushed by the nocurrentpoint handler).  This too
will be fixed in 2.5.1.

Thank you for bringing these bugs -- both of which have been in
Ghostscript for a long time -- to my attention.

L. Peter Deutsch :: Aladdin Enterprises :: P.O. box 60264, Palo Alto, CA 94306
ghost@aladdin.com, ...decwrl!aladdin!ghost ; voice 415-322-0103 ; fax 322-1734
    "Implementation is the sincerest form of flattery."    new--->^^^^^^^^^^^^