gmp build fail on fedora 9

[dbk]

Hi all,

I'm getting a strange configure problem building 3.02 on fedora 9
running on X86_64 system;

First I get;
****************************************************
****************************************************
GCC Version
gcc -v
Using built-in specs.
Target: x86_64-redhat-linux
Configured with: ../configure --prefix=/usr --mandir=/usr/share/man --
infodir=/usr/share/info --with-bugurl=http://bugzilla.redhat.com/
bugzilla --enable-bootstrap --enable-shared --enable-threads=posix --
enable-checking=release --with-system-zlib --enable-__cxa_atexit --
disable-libunwind-exceptions --enable-languages=c,c++,objc,obj-c+
+,java,fortran,ada --enable-java-awt=gtk --disable-dssi --enable-
plugin --with-java-home=/usr/lib/jvm/java-1.5.0-gcj-1.5.0.0/jre --
enable-libgcj-multifile --enable-java-maintainer-mode --with-ecj-jar=/
usr/share/java/eclipse-ecj.jar --disable-libjava-multilib --with-
cpu=generic --build=x86_64-redhat-linux
Thread model: posix
gcc version 4.3.0 20080428 (Red Hat 4.3.0-8) (GCC)
****************************************************

which is OK, not great but OK, then when the gmp build starts I get;


checking build system type... pentium3-unknown-linux-gnu
checking host system type... pentium3-unknown-linux-gnu
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of
Makefiles... no
checking ABI=32
checking compiler gcc -m32 -O2 -fomit-frame-pointer ... no
checking compiler gcc -O2 -fomit-frame-pointer ... yes
checking compiler gcc -O2 -fomit-frame-pointer has sizeof(long)==4...
no
checking compiler icc -no-gcc ... no
configure: error: could not find a working compiler, see config.log
for details

this doesn't make sense to me?? why ABI=32 and gcc is working fine. I
tried to build gmp from the the sage shell but get the same result.

Also, it seems like configure should have cpu to core2 since I'm
running on a Xeon X5482 system??

Anyone have any ideas.

Thanks
-Don

[mabshoff]

On Jun 21, 7:35 pm, dbk <d...@dnc-arts.com> wrote:
> Hi all,

Hi Don,

> I'm getting a strange configure problem building 3.02 on fedora 9
> running on X86_64 system;

The problem is not FC9 specific, see below for more info.

Yes, this is a known issue with gmp since it uses cpuid and not the
extended cpuid call, so certain new Core2 cpus (like the one you have)
is identified as a P3. We are about to switch to our own gmp for
called mpir where we fixed that issue. The plan is to have MPIR in the
next Sage release 3.1.0, but that depends on a couple things we need
to do this week. If MPIR we will likely back port the fix into the gmp
4.2.2 release.

The whole discussion can be found at

http://groups.google.com/group/mpir-devel/t/c31f89ed1683d383

As a short term solution you can force the host to be something
specific, but that slows down performance by a lot. Let me know and I
can dig out the exact info for that.

> Anyone have any ideas.
>
> Thanks
> -Don

Cheers,

Michael

[dbk]

Hi Michael,

Thanks for the discussion link. I'll play around a little with the
patch and see if I can do anything with config.guess.
Looks like a pretty tricky build! If I come up with anything useful
I'll be sure to let you know.

Take care
-Don

[Rex]

On Jun 21, 9:35 pm, dbk <d...@dnc-arts.com> wrote:
> I'm getting a strange configure problem building 3.02 on fedora 9

One unfortunate side effect of bundling dependencies instead of
allowing the
use those provided by the host os/system. hint hint, grumble,
grumble.

-- Rex

[William Stein]

Heh, we do provide a version of Sage without any
bundled dependencies:

http://sagemath.org/packages/standard/sage-3.0.3.spkg

(The spkg is just a tar.bz2.) Just extract the above,
get all 70-80 dependencies installed, then do

python setup.py install

with the above, and you're good to go.

Tim Abbott took up the above challenge, and soon Sage
will be in Debian...

-- William

[Rex]

On Jun 23, 10:08 am, "William Stein" <wst...@gmail.com> wrote:

> Heh, we do provide a version of Sage without any
> bundled dependencies:
>
> http://sagemath.org/packages/standard/sage-3.0.3.spkg
>
> (The spkg is just a tar.bz2.) Just extract the above,
> get all 70-80 dependencies installed, then do
>
> python setup.py install
>
> with the above, and you're good to go.
>
> Tim Abbott took up the above challenge, and soon Sage
> will be in Debian...

Many thanks to all for those efforts, as it is well-spent groundwork
for
including sage into other distros as well.

-- Rex

[William Stein]

Definitely. I think Tim has done a heroic job, with help
from Michael Abshoff and others, and it will help greatly
getting Sage into other distros. I will be very very happy
when this happens.

-- William