aed@netcom.UUCP (Andrew Davidson) (12/11/90)
I have been trying to build gcc on sco unix for some time know. Chip gave me a set fof patches but I still have the same reacuring problem 5% more err.lis7 rcc -c -g -I. -I. -I./config fold-const.c rcc -c -g -I. -I. -I./config rtl.c "rtl.c", line 175: warning: illegal pointer combination, op = "rtl.c", line 243: warning: illegal pointer combination, op != "rtl.c", line 249: warning: illegal pointer combination, op != "rtl.c", line 356: warning: illegal pointer combination, op != "rtl.c", line 392: warning: illegal pointer combination, op != "rtl.c", line 445: warning: illegal pointer combination, op != "rtl.c", line 660: warning: illegal pointer combination, op = "rtl.c", line 679: warnng: illegal pointer combination, op : "rtl.c", line 679: illegal types in : *** Error code 1 Stop. I would appriciate any ideas. (it is hard to debug someone elses code) thanks. ps it has nothing to do with larry walls patch program andy
drmorris@athena.mit.edu (David R Morrison) (12/11/90)
In article <18496@netcom.UUCP> aed@netcom.UUCP (Andrew Davidson) writes:
I have been trying to build gcc on sco unix for some time know. Chip gave me
a set fof patches but I still have the same reacuring problem
5% more err.lis7
rcc -c -g -I. -I. -I./config fold-const.c
rcc -c -g -I. -I. -I./config rtl.c
*warnings*
"rtl.c", line 679: illegal types in :
*** Error code 1
Stop.
I would appriciate any ideas. (it is hard to debug someone elses code)
thanks.
I had this exact problem last summer. What fixed it:
(this is from memory, so check me)
----------------------------------------------------------------------
XVEC (return_rtx, i) = (list_counter
? rtvec_alloc (list_counter)
: NULL);
----------------------------------------------------------------------
if (list_counter)
{ XVEC (return_rtx, i) = rtvec_alloc (list_counter) }
else
{ XVEC (return_rtx, i) = NULL}
----------------------------------------------------------------------
I assume it is an rcc bug. This should be put in the SCO patches, if it
isn't there already...
Dave Morrison
sef@kithrup.COM (Sean Eric Fagan) (12/11/90)
In article <DRMORRIS.90Dec10233201@hal-2000.mit.edu> drmorris@athena.mit.edu (David R Morrison) writes: >I assume it is an rcc bug. This should be put in the SCO patches, if it >isn't there already... I didn't remember this until I saw the code. I am fairly certain this is because one of the pointers is a 'void*', and I don't believe that rcc likes 'void*'. Either that, or the left-hand side of the ':?' expression was a type it didn't like (possibly an enum?). It's been a *long* time since I've compiled gcc, so I don't remember completely. Sorry... I mostly used /bin/cc to compile gcc, and resorted to rcc a couple of times when cc failed. If you use rcc, remember to use define M_UNIX, M_COFF, and M_I386 (i386 and unix should already be defined, and set to 1, but you may want to stick them in there, just in case). -- Sean Eric Fagan | "I made the universe, but please don't blame me for it; sef@kithrup.COM | I had a bellyache at the time." -----------------+ -- The Turtle (Stephen King, _It_) Any opinions expressed are my own, and generally unpopular with others.
chip@tct.uucp (Chip Salzenberg) (12/15/90)
According to drmorris@athena.mit.edu (David R Morrison): > I have been trying to build gcc on sco unix for some time know. Chip gave > me a set of patches but I still have the same reacuring problem My patches included fixes for the SCO /usr/include directory. The compile errors shown are usually due to an incorrect definition of NULL -- the AT&T compiler ("rcc") can't handle "((void *)0)". An aside -- Can you believe that SCO'S OWN HEADERS have THREE different definitions for NULL? Wait, what am I saying? Of course you can believe it... :-( :-( My patches for /usr/include are one of the files in my GCC patch kit for SCO UNIX. The filename is "SCO.include". -- Chip Salzenberg at Teltronics/TCT <chip@tct.uucp>, <uunet!pdn!tct!chip> "Please don't send me any more of yer scandalous email, Mr. Salzenberg..." -- Bruce Becker