home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: biz.sco.general
- Path: sparky!uunet!spool.mu.edu!umn.edu!csus.edu!netcom.com!vancleef
- From: vancleef@netcom.com (Henry van Cleef)
- Subject: Re: compiling GNU programs
- Message-ID: <1993Jan2.020535.12279@netcom.com>
- Organization: Netcom--posted from Andover, Mass.
- References: <9212302046.aa12907@ohriman.ohriman.mi.org>
- Date: Sat, 2 Jan 1993 02:05:35 GMT
- Lines: 55
-
- In article <9212302046.aa12907@ohriman.ohriman.mi.org> jep@ohriman.mi.org writes:
- > I've got a few problems compiling GNU software which I'm hoping
- >someone here can help with.
- > I'm running gcc 2.3.3 (which, by the way, appears to compile without
- >problems using the instructions for SCO). I've had the same problems
- >with 2.2.2 and 2.3.2 in the past, which I compiled using different
- >sets of instructions and patches provided by people on this list.
- > When I try to compile gdb 4.7, everything appears to go smoothly
- >until it gets to "readline.c". There I get massive errors; parse
- >errors originating with /usr/include/sys/stream.h, including:
-
- You will have to be specific about what version of SCO Unix you are
- using. YOur comments imply 3.2.4, and I have spent the last few days
- living with GNU's attitudes towards SCO's 3.2.4 setup.
-
- The gdb readline.c will compile if you set the CC line in the makefile
- to gcc -traditional. If you still have trouble, try forcing
- -D_NO-PROTOTYPE in the CFLAGS.
-
- All else fails, compile it with the stock cc. If you have an early
- version of the 3.2.4 devsys, get SLS lng353a from sosco---this is the
- version of the stock cc I am using.
-
- The problems I have had, and that appear to be yours from your
- description, stem from the header #ifdefs for __STDC__, _X_OPENSOURCE
- and _POSIX_SOURCE. GCC won't take full function prototypes under
- -traditional, and SCO has succeeded in preventing the errno.h
- assignments in some cases. On top of which, various GNU things claim to
- be ansi or posix, but aren't, in terms of SCO's headers. A brute force
- -U__STDC__ without the -traditional flag seems to make a lot of problems
- go away with GCC.
-
- My config is 3.2 vee four, GCC 2.3.3, GDB 4.7. I have not yet taken a
- whack at libg++, although have notes on what it took to make that happen
- (plenty) on a 3.2.0 system with GCC-2.3.2.
-
- Don't be deluded by the idea that GNU software is either clean or
- consistent with other GNU things. -Di386, -U__STDC__, and _NO_PROTOTYPE
- seem to make things work. Read your headers, and learn how to read the
- #ifdefs. Those are what you are going to compile with.
-
- As far as the finxinc.sco script that comes with GCC 2.3.3, I have not
- run it, and it looks to me like a horrible hack.
-
- So far as standards are concerned, it is much more difficult to port
- code that claims to meet standards these days, and I am very tired of
- the stuff around "Standard C." Shutting off of all the error numbers in
- sys/errno.h is the living end, and I "fixed" that problem on my system.
-
-
- --
- Hank van Cleef vancleef@netcom.com vancleef@tmn.com Andover, Mass.
- Unix systems consultant---kernel, device drivers, networking, X11
- SysV, BSD, Sun, Ultrix, AIX, SCO. Porting a specialty.
- The Union Institute History of Science
-