[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: depcomp problem [Fwd: Trying to compile latest CVS on old SC Ounix
From: |
Tom Tromey |
Subject: |
Re: depcomp problem [Fwd: Trying to compile latest CVS on old SC Ounixware 2] |
Date: |
18 Jul 2001 00:09:50 -0600 |
>>>>> "Steve" == Cameron, Steve <Steve.Cameron@COMPAQ.com> writes:
Steve> The little "compile" script complains that there is no "-o" option.
Steve> (If there were a -o option, the SCO compiler would complain, so this
Steve> is probably intentional.)
Steve> I've gone over this on the cvs-bug@gnu.org, but it seems to be
Steve> an automake problem and Derek Price, one of the CVS
Steve> maintainers, suggested I try here. Previously there was some
Steve> discussion here on automake@gnu.org about this problem which
Steve> resulted in the suggestion to add AM_PROG_CC_C_O to into
Steve> configure.in, which appears to have been done, but resulted in
Steve> the above.
I looked into this tonight. I couldn't reproduce your failure, but
I'm definitely seeing weird things happen. I'll be checking in a test
case soon. This means the problem will be fixed before the next
release.
In particular I noticed:
- We get different results if we run AM_PROG_CC_C_O before AC_PROG_CC.
Maybe this one is just an autoconf (2.13) bug. Why doesn't
AC_PROG_CC_C_O require AC_PROG_CC?
- AM_PROG_CC_C_O changes `CC'. This is bad enough, but it also
changes it like this:
CC="\$(top_srcdir)/compile $CC"
This means if your configure script tries to use $CC after
AM_PROG_CC_C_O then it will fail. Oops!
- In my case, a lock directory with the name of the expected object is
left lying around.
I think all these are fixable (though the second one perhaps hackily).
Tom