This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

`make distclean' broken (was failure to bootstap on sparc-sun-solaris2.7)


>>>>> "Bernd" == Bernd Schmidt <bernds@redhat.com> writes:

Bernd> On Thu, 18 Jan 2001, Martin Buchholz wrote:
>> gcc seems to want to do glibc-related things, even though this is a
>> fairly standard Solaris system.  No glibc anywhere to be found.

Bernd> Does this work with 2.95.2?  If it does, can you try backing out the
Bernd> changes in the gcc/fixinc directory?

I did an experiment that confirmed my suspicion.

Both native configurations, i686-pc-linux-gnu and
sparc-sun-solaris2.7, bootstrap just fine when executed in a pristine
directory from untarring the gcc-2.95.3.test2 distribution tarball
with objdir == srcdir.

However, if you take again such a pristine directory, and do

(on linux) make distclean; ./configure && make bootstrap
(on solaris) make distclean; ./configure && make bootstrap

then the solaris build fails, presumably because the `make distclean'
did not remove all the files it should have.

The exact build command was:

( make distclean; ./configure --prefix=/usr/local --disable-nls --enable-languages='c,c++' && make bootstrap ) >>../linaris.log 2>&1

For every supported platform, you should not test just that the make
bootstrap works, but that after it does, `make clean' results in a
directory identical to the pristine state after first untarring.  The
tool to use is `diff -r --brief'.

Build logs are available.

gcc-3.0 should also be checked in the same fashion.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]