Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!cmcl2!rna!dan
From: dan@rna.UUCP (Dan Ts'o)
Newsgroups: comp.graphics,comp.bugs.sys5
Subject: Porting System V GPS graphics to the SUN 3...
Message-ID: <656@rna.UUCP>
Date: Sun, 20-Sep-87 23:57:25 EDT
Article-I.D.: rna.656
Posted: Sun Sep 20 23:57:25 1987
Date-Received: Tue, 22-Sep-87 01:23:09 EDT
Organization: Rockefeller Neurobiology
Lines: 39
Xref: mnetor comp.graphics:1187 comp.bugs.sys5:243


	I started porting GPS to SUN 3's. One must declare the large auto
arrays in plot.c and subset.c as static. I still don't know why the SUN's
should have the restriction of limited local variable space (and I believe they
shouldn't: its a bug in my book).
	However, that's not all. Nearly every GPS filter does a NULL
reference when cycling through argv, with code like this:

	if(strcmp(*++argv,"-")==0 || !argf ) fdi=stdin;

where *++argv is often NULL, especially when invoking the filter without
any arguments. I did the quickest mod to fix this:

	if((*++argv && strcmp(*argv,"-")==0) || !argf ) fdi=stdin;

	Also several bits of the GPS code are byte-order dependent. Most
are flagged by the conditional:

#if u3b | u370

	to which I added:

#if u3b | u370 | mc68000

	In addition, a few programs require <sys/termio.h>. The easiest
thing to do is add a -I/usr/5include argument in their cca scripts. I
believe the programs that look for this are part of the tek4000.d stuff.

	I`m sure I haven't uncovered all the problems (e.g. I haven't ported
GED yet) but at least now I can use most of the common GPS filters and plot
the results using plot(GPS) and td on a SUN 3 using tektool.

	Too bad GPS is considered obsolete (in preference to what ?) by
AT&T. Its a nice package. BTW, I have a few more filters now, including an
FFT filter.

P.S. Last minute update: I did manage to port ged. It appears to run correctly
	using tektool, input cursors and all.