🚀 go-pugleaf

	There  seems  to be a bug  in  the  implementation  of the pathforall
operator [for both gs2.4.1 and gs2.5].  According  to the PostScript Language
Reference  Manual, second  edition  : "If any  of  the  procedures change the
current path, such changes do not alter the behavior of pathforall" [p462].

	That is not the case right now. If the  procedures invoked during the
execution of the pathforall create a  path that  is longer than  the original
one, the two paths get mixed up with undesirable results. The following patch
seems to solve the problem when applied to file zpath2.c [from the source for
ghostscript 2.5] :

31,34d30
< #include "gxfixed.h"
< #include "gxmatrix.h"
< #include "gzstate.h"
< 
86d81
< 	gx_path_share(igs->path);

	However, not being a ghostscript expert [it was the first time I even
looked at the  code],  I am probably  overlooking something (I'm not sure  if
this path ever gets freed for one thing). Would anyone care to confirm/infirm
this, and fix the patch if necessary?

-- 
/***************************************************************************\
|* Patrice Belleville 					 patrice@cs.sfu.ca *|
|*   --- Simon Fraser University is not responsible for my mistakes. ---   *|
\***************************************************************************/