libsupc++ changes broke libjava build??

[Phil Edwards]

On Wed, Mar 28, 2001 at 10:21:41PM +1200, Bryce McKinlay wrote:
> This patch seems to be the most likely culprit:
> 
> 2001-03-27  Alexandre Oliva  <aoliva@redhat.com>
> 
>         * libsupc++/Makefile.am (CXXLINK): Use CXX again, and choose
>         CXX tag explicitly.
>         * src/Makefile.am (CXXLINK): Likewise.
>         * libsupc++/Makefile.in, src/Makefile.in: Rebuilt.

Reverting this patch alone doesn't help; the lib-v3 testsuite ends up with
failures like

libtool: link: only absolute run-paths are allowed

/home/pme/build/install-2001-03-28/bin/g++ -ggdb3 -DDEBUG_ASSERT
-I/home/pme/src/unified/libstdc++-v3/testsuite -o ./header_cassert.sh-exe
/home/pme/build/install-2001-03-28/lib/libstdc++.so -lm -lm
-lm -lm -Wl,--rpath -Wl,/home/pme/build/install-2001-03-28/lib
-Wl,--rpath -Wl,/home/pme/build/install-2001-03-28/lib -Wl,--rpath
-Wl,/home/pme/src/unified/libstdc++-v3/testsuite/17_intro/header_cassert.cc
/usr/lib/crt1.o: In function `_start':
/usr/lib/crt1.o(.text+0x18): undefined reference to `main'
collect2: ld returned 1 exit status


Note the "-Wl,--rpath -Wl,/path/to/source.cc" at the end.  Whoops.


Phil

-- 
pedwards at disaster dot jaj dot com  |  pme at sources dot redhat dot com
devphil at several other less interesting addresses in various dot domains
The gods do not protect fools.  Fools are protected by more capable fools.