Installation of Sage 8.7 crashes in mpc-1.1.0

[Pepijn Wissing]

Dear reader,

I have been attempting to get Sage to run on a computational cluster, which is operated by a Debian (to which I am quite unfamiliar) system, via the following terminal commands:

module load Python/2.7.15-foss-2018b  # i was told to include this as it would be "better" since several Python packages would have to be run with Sage
wget https://mirror.koddos.net/sagemath/src/sage-8.7.tar.gz
tar -zxf sage-8.7.tar.gz
cd sage-8.7
make

However, as the subject line suggests, it crashes after about ten minutes, during the installation of mpc-1.1.0. I have attached the log files to this post. Would anyone be able to assist me?

Kind regards,

Pepijn

[Vincent Delecroix]

Dear Pepijn,

Your config.log mentions that you ran

$ ./configure --prefix=/home/pwissing/sage-8.7/local
--libdir=/home/pwissing/sage-8.7/local/lib
--with-gmp=/home/pwissing/sage-8.7/local
--with-mpfr=/home/pwissing/sage-8.7/local

which is not what you described in your mail. Is there an omission in
your mail or a mistake in the config.log? The above command line looks
very wrong.

Best
Vincent

[François Bissey]

Hi,

configure:13910: gcc -o conftest -m64 -O2 -march=corei7-avx -mtune=corei7-avx -g -I/home/pwissing/sage-8.7/local/include -I/home/pwissing/sage-8.7/local/include -L/home/pwissing/sage-8.7/local/lib -L/home/pwissing/sage-8.7/local/lib -L/home/pwissing/sage-8.7/local/lib -Wl,-rpath,/home/pwissing/sage-8.7/local/lib conftest.c -lmpfr -lgmp -lm >&5
/home/pwissing/sage-8.7/local/lib/libmpfr.so: error: undefined reference to '__gmpn_com'
collect2: error: ld returned 1 exit status

Hum. Something is wrong with mpfr. Could you provide the following
readily -d /home/pwissing/sage-8.7/local/lib/libmpfr.so
and
ldd -r /home/pwissing/sage-8.7/local/lib/libmpfr.so

François

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/0635c4a9-2c0e-452e-8e5b-26e0e52f30c0%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
> <install.log><mpc-1.1.0.log><config.log>

[Jeroen Demeyer]

On 2019-05-09 10:55, Vincent Delecroix wrote:
> Dear Pepijn,
>
> Your config.log mentions that you ran
>
> $ ./configure --prefix=/home/pwissing/sage-8.7/local
> --libdir=/home/pwissing/sage-8.7/local/lib
> --with-gmp=/home/pwissing/sage-8.7/local
> --with-mpfr=/home/pwissing/sage-8.7/local

That's the config.log of mpc, which looks OK.

[Jeroen Demeyer]

I think that your environment is too messed up. Why do you have all
those environment variables like $CPATH and $LD_LIBRARY_PATH set? That's
probably what's causing the breakage. Can you try to build in a clean
environment, without setting any such environment variables?

[Dima Pasechnik]

yeah, this seems to indicate that it's a conflict with Conda installation.

Not sure about the meaning of module "load Python/2.7.15-foss-2018b".
Note that Sage will built its own copy of Python,
so there is totally no need for any "extra Python packages" to start with.

mvg
Dima

>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/5CD3EF0D.7090000%40UGent.be.

[Pepijn Wissing]

I'm attempting to get Sage installed on the SURFsara computational cluster; I can only assume any such variables are set by the load management system (SLURM) that distributes jobs to batch nodes, though I'm not too familiar with the exact inner workings of the system. As said, I have only entered the commands above, after connecting to a login node. I'll try to get a clean environment, to be continued... ;-) 

Op donderdag 9 mei 2019 11:12:48 UTC+2 schreef Jeroen Demeyer:

[Pepijn Wissing]

Right, so thanks to your answers I have run just the installer, without loading the python module, which has in part solved my problem; it has succesfully installed mpc. However, it now fails in the installation of giac. I have attached the new logs. Any further suggestions as to what the problem might be? 

Full install log via a dropbox link: https://www.dropbox.com/s/f50vdhxeq4g521p/install.log?dl=0 

 

[Pepijn Wissing]

As far as I can tell, it fails when it's attempting to reach something called "cocoa", about which it says 

[giac-1.5.0.37.p0] checking CoCoA/io.H usability... no
[giac-1.5.0.37.p0] checking CoCoA/io.H presence... no
[giac-1.5.0.37.p0] checking for CoCoA/io.H... no

earlier. Do you think something is missing or is there a bigger issue at play?

Op donderdag 9 mei 2019 16:25:39 UTC+2 schreef Pepijn Wissing:

[E. Madison Bray]

On Thu, May 9, 2019 at 4:35 PM Pepijn Wissing <pepijn....@gmail.com> wrote:
>
> As far as I can tell, it fails when it's attempting to reach something called "cocoa", about which it says
> [giac-1.5.0.37.p0] checking CoCoA/io.H usability... no
> [giac-1.5.0.37.p0] checking CoCoA/io.H presence... no
> [giac-1.5.0.37.p0] checking for CoCoA/io.H... no
> earlier. Do you think something is missing or is there a bigger issue at play?

I don't think this is relevant.

What's strange, in your log, is that it errored out on this particular
step with no specific error message of use, because for some reason
libtool had it run output redirected to /dev/null (?). So you might
try manually re-running the command it failed on. First activate the
sage environment:

$ ./sage -sh

Then cd to the build directory where giac was being built:

$ cd /home/pwissing/sage-8.7/local/var/tmp/sage/build/giac-1.5.0.37.p0/src/src

Finally, manually run the command that failed:

$ g++ -std=gnu++11 -DHAVE_CONFIG_H -I. -I.. -DIN_GIAC -I. -I.. -I.
-I.. -I/home/pwissing/sage-8.7/local/include -g -O2
-fno-strict-aliasing -DGIAC_GENERIC_CONSTANTS -MT cocoa.lo -MD -MP -MF
.deps/cocoa.Tpo -c cocoa.cc -o cocoa.o

and see if that gives anything more useful.

This seems kind of random though so I'd suspect something
environmental (out of disk space?).

> Op donderdag 9 mei 2019 16:25:39 UTC+2 schreef Pepijn Wissing:
>>>
>>> Right, so thanks to your answers I have run just the installer, without loading the python module, which has in part solved my problem; it has succesfully installed mpc. However, it now fails in the installation of giac. I have attached the new logs. Any further suggestions as to what the problem might be?
>>
>>
>> Full install log via a dropbox link: https://www.dropbox.com/s/f50vdhxeq4g521p/install.log?dl=0
>>
>>
>

> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To post to this group, send email to sage-...@googlegroups.com.
> Visit this group at https://groups.google.com/group/sage-devel.

> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/5f1c7377-08c6-4b42-8584-3981a2ca0614%40googlegroups.com.

[Pepijn Wissing]

Thanks to all that have replied, I have (for the moment) been able to get some version of Sage running by installing a pre-compiled version, specifically, this one: https://mirror.koddos.net/sagemath/linux/64bit/sage-8.7-Debian_GNU_Linux_9-x86_64.tar.bz2

Op donderdag 9 mei 2019 10:46:11 UTC+2 schreef Pepijn Wissing: