Those funny options are for the native SGI C compiler. I have never been able
to compile any versions of Postgresql using gcc under Irix. Switching to the
native cc, I've had very few problems (including no problems with 7.0.3). The
list of numeric options turns off a number of warning messages from the native
cc; they tend to be so numerous that real errors are tough to spot. The
particular list begins with the suggested list from SGI for compiling
"freeware" which has been tuned to compile with gcc
(http://freeware.sgi.com/shared/howto.html). I modified the list for postgresql
which generated its own list of errors.
Perhaps a way is needed to differentiate between the two compilers; Some kind of
a test within the Makefile which looks not only at the machine type?
The quote from SGI is:
> Catching other problems
>
> The SGI cc is verbose and pedantic. The problem is that when you get way too may warnings you start ignoring the real
ones.
>
> To make the SGI compiler more like gcc in terms of warnings you may use something like:
>
> NOWARN = -woff 1009,1014,1048,1110,1116,1185,1188,1204,1230,1233 \
> -Wl,-woff,85,-woff,84
> $(CC) $(NOWARN) ...
>
> "Marek Bobrowski" writes:
> > gcc -I../../../include -I../../../backend -U_NO_XOPEN4 -woff 1164,1171,1185,1195,1552 -Wl,-woff,15 -Wl,-w,84
-Wall
> -Wmissing-prototypes -Wmissing-declarations -I../.. -c indextuple.c -o indextuple.o
> > gcc: 1164,1171,1185,1195,1552: No such file or directory
> > gcc: unrecognized option `-woff'
> > gmake[3]: *** [indextuple.o] Error 1
>
> Wow. Where did all that cruft come from? The "-U_NO_XOPEN4 -woff
> 1164,1171,1185,1195,1552 -Wl,-woff,15 -Wl,-w,84" should not be there.
>
> Do you perhaps have CFLAGS or CPPFLAGS set in your environment?
>
> regards, tom lane
>
Hope this helps,
Mark Dalphin (mdalphin@amgen.com)