*BSD News Article 4884


Return to BSD News archive

Newsgroups: comp.unix.bsd
Path: sserve!manuel!munnari.oz.au!hp9000.csc.cuhk.hk!uakari.primate.wisc.edu!zaphod.mps.ohio-state.edu!rpi!usc!venice!gumby.dsd.trw.com!eel.dsd.trw.com!gottloeb
From: gottloeb@eel.dsd.trw.com (Jeffrey R. Gottloeb)
Subject: Re: ghostscript under 386BSD?
Message-ID: <1992Sep11.184107.11696@gumby.dsd.trw.com>
Sender: news@gumby.dsd.trw.com
Organization: TRW, INC.
References: <s-vn48f.alm@netcom.com>
Date: Fri, 11 Sep 1992 18:41:07 GMT
Lines: 30

In article <s-vn48f.alm@netcom.com> alm@netcom.com (Andrew Moore) writes:
>Ghostscript is part of the /etc distribution of 386bsd. I rebuilt it
>using the default makefile (for epson device). But when invoked it dies
>with diagnostic:
>
>Initializing... Zget t_string 256 >= 256
>Unrecoverable error: rangecheck in get
>Operand stack:
>    256 [...]
>
>Before I attempt to wade through the code, I was hoping someone might
>have fixed this.  Any ideas?  Anyone else successfully using ghostscript
>under 386bsd?  Thanks.
>-Andrew Moore <alm@netcom.com>


I noticed this problem too.  I believe is a compiler problem.  If
ghostscript is compiled without optimization and with debug, floating
point exceptions are generated.  I have tracked the problem as far as
when the setscreen command in gs_init.ps is being interpreted.

There is a notation in the PROBLEMS file for gcc that the compiler
generates bad code on a 386 for certain cases of floats.

I am trying to build gcc-2.2.2 on my system as we speak.  Perhaps this
will clear up the problem when i recompile ghostscript with it.


Jeff Gottloeb
gottloeb@gumby.dsd.trw.com