ghost@SUN.COM (L. Peter Deutsch) (03/27/89)
> From: uunet.uu.net!mcvax!kth!draken!liuida!prosys!ath (Anders Thulin) > Organization: Programsystem AB, Linkoping, SWEDEN > Subject: Re: Bugs ... > Message-Id: <403@helios.prosys.se> > > * Invoking gs with an unknown option (-p) produces an 'Unknown switch > -p' message and a core dump. The core dump seems unnecessary. > gsmain(gsmain.c) calls abort() rather than exit(). I agree, this is silly. On the Ghostscript development system, abort() doesn't produce a core dump, and I didn't know it did this on Unix systems. I've changed all occurrences of abort() to exit(1). > * There is no statusdict. Hence some 'investigative' programs will > fail. The one I was using was PSDoctor 1.0.0, recently distributed > on Usenet. Adding some definitions in ghost.ps fixed the problem. > Note: my ideas of printer names, etc. might not be yours. I don't have much enthusiasm for catering to PostScript Programs that depend on features specific to the LaserWriter. What I've done is made a separate file, statusd.ps, that contains a version of your suggested statusdict definitions. I've changed ghost.ps so it creates an empty statusdict in systemdict: if you want the full set of definitions, you can use a script that reads in statusd.ps when you fire up Ghostscript, or you can add (statusd.ps) run to your version of ghost.ps. > * PSDoctor also showed a problem with 'usertime'. The call to clock() > in gs_usertime(gsmisc.c) returns microseconds rather than > milliseconds as wanted. Thanks for the fix. > * Due to the redefinition of findfont in gfonts.ps it was impossible > to create user defined fonts (see red book, p. 101-102). Changing the > redefinition at bit fixed the problem. Thanks for the fix. > No patches for these: > > * Halftone screens appear to be aligned with user coordinates rather > than device coordinates. See the star example on page 51 in the red > book. I assume you mean p. 51 of the blue book, yes? This must be a bug in the X back end: it works just fine on the PC. I remember having a lot of trouble with halftones under X (one of several X facilities that I tried using and that I never figured out how to make work). If someone else can track this down, I'd be more than happy to install a fix. > * The lineofcircles example in the blue book (p. 69-70) never > terminates. Debug printouts of 'xpos', 'pagewidth', and 'xpos > pagewidth le' show that although xpos is increased beyond pagewidth > (i.e. 612), the expression 'xpos pagewidth le' doesn't become false > when x > pagewidth. Someone else reported this problem too, and I'm quite puzzled by it. I ran the example on a Sun-4 using the 1.2 distribution sources, and it terminated properly. The definition of the 'le' operator is in file zrelbit.c, procedures zle and obj_compare: if you could take a little time to analyze the failure, I'd be very grateful. > My suggested patches to the 1.2 distribution: Your patches are fine, the ones I'll be distributing are essentially the same. Thanks for helping improve Ghostscript! - P. D. aladdin!ghost@parcplace.com ...{sun!}parcplace!aladdin!ghost P. O. box 60264, Palo Alto, CA 94306