Many tests failures

[Philippe Delavalade]

Hi all.

I just installed sage-10.7. Installation seemed to be correct.

When running
make ptestlong
I got many errors and failures. I don't understand what is going wrong.

I join config.log, install.log and test.log (this one is very long).

I'm using linux, slint 15.0 (based on slackware 15.0).

Thanks for help.

--
Philippe

[Dima Pasechnik]

Hi,
the errors in your test.log all appear to come from a broken
matplotlib installation.
A typical error message is

ImportError: /home/math/sage-10.7/local/var/lib/sage/venv-python3.12.5/lib/python3.12/site-packages/matplotlib/_path.cpython-312-x86_64-linux-gnu.so:
undefined symbol: _ZN3agg10curve3_div4initEdddddd

("undefined symbol" varies, but it's always in
"_path.cpython-312-x86_64-linux-gnu.so")

Can you post the output of

ldd /home/math/sage-10.7/local/var/lib/sage/venv-python3.12.5/lib/python3.12/site-packages/matplotlib/_path.cpython-312-x86_64-linux-gnu.so

?

It's certain C++ function, curve3_div, as you can see by running the demangler:

c++filt _ZN3agg10curve3_div4initEdddddd
agg::curve3_div::init(double, double, double, double, double, double)

Apparently it's from https://matplotlib.org/stable/api/backend_agg_api.html

Can you post the matplotlib install log, in logs/pkgs/matplotlib*.log ?

HTH
Dima

On August 21, 2025 4:23:42 AM CDT, Philippe Delavalade

[Philippe Delavalade]

Hi Dima and thanks for your answer.

I noticed that it was always the same kind of error.


Le jeudi 21 août à 16:44, Dima Pasechnik a écrit :

> Hi,
> the errors in your test.log all appear to come from a broken
> matplotlib installation.
> A typical error message is
>
> ImportError: /home/math/sage-10.7/local/var/lib/sage/venv-python3.12.5/lib/python3.12/site-packages/matplotlib/_path.cpython-312-x86_64-linux-gnu.so:
> undefined symbol: _ZN3agg10curve3_div4initEdddddd
>
> ("undefined symbol" varies, but it's always in
> "_path.cpython-312-x86_64-linux-gnu.so")
>
> Can you post the output of
>
> ldd /home/math/sage-10.7/local/var/lib/sage/venv-python3.12.5/lib/python3.12/site-packages/matplotlib/_path.cpython-312-x86_64-linux-gnu.so

Output is
linux-vdso.so.1 (0x00007f7aff601000)
libstdc++.so.6 => /usr/lib64/libstdc++.so.6 (0x00007f7aff200000)
libm.so.6 => /lib64/libm.so.6 (0x00007f7aff444000)
libgcc_s.so.1 => /usr/lib64/libgcc_s.so.1 (0x00007f7aff429000)
libc.so.6 => /lib64/libc.so.6 (0x00007f7aff021000)
/lib64/ld-linux-x86-64.so.2 (0x00007f7aff603000)

> It's certain C++ function, curve3_div, as you can see by running the demangler:
>
> c++filt _ZN3agg10curve3_div4initEdddddd
> agg::curve3_div::init(double, double, double, double, double, double)
>
> Apparently it's from https://matplotlib.org/stable/api/backend_agg_api.html
>
> Can you post the matplotlib install log, in logs/pkgs/matplotlib*.log ?

I join it.

Cheers.

--
Philippe

>
> HTH
> Dima
>
> On August 21, 2025 4:23:42 AM CDT, Philippe Delavalade
> <philippe....@gmail.com> wrote:
> >Hi all.
> >
> >I just installed sage-10.7. Installation seemed to be correct.
> >
> >When running
> >make ptestlong
> >I got many errors and failures. I don't understand what is going wrong.
> >
> >I join config.log, install.log and test.log (this one is very long).
> >
> >I'm using linux, slint 15.0 (based on slackware 15.0).
> >
> >Thanks for help.
> >
>

> --
> You received this message because you are subscribed to the Google Groups "sage-support" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-support...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/sage-support/CAAWYfq3%2BLYS%2B-Uu3PUx%2BvG%2BA2AryjBEh2txCrouDz5tJjCwGrA%40mail.gmail.com.

[Dima Pasechnik]

Thanks for the log; one place there looks suspicious, and probably
results in your errors

[spkg-install] [9/30] Linking static target extern/agg24-svn/libagg.a
[spkg-install] ar:
extern/agg24-svn/libagg.a.p/src_agg_bezier_arc.cpp.o: plugin needed to
handle lto object

So it's probably ar which needs to be replaced by gnu-ar, don't know
how, though:
https://gcc.gnu.org/legacy-ml/gcc-help/2016-01/msg00061.html
If you can manage this, and rebuild, fine. If not,
can you try merging the branch pull/3/head of
https://github.com/cxzhong/sage pull/3/head and see if it builds.
This is an update of matplotlib and its deps, hopefully it will fix this all.

On Thu, Aug 21, 2025 at 12:22 PM Philippe Delavalade

> To view this discussion visit https://groups.google.com/d/msgid/sage-support/aKdQ8QwGjhuuzEJ2%40messier31.example.messier31.

[Dima Pasechnik]

On Thu, Aug 21, 2025 at 1:19 PM Dima Pasechnik <dim...@gmail.com> wrote:
>
> Thanks for the log; one place there looks suspicious, and probably
> results in your errors
>
> [spkg-install] [9/30] Linking static target extern/agg24-svn/libagg.a
> [spkg-install] ar:
> extern/agg24-svn/libagg.a.p/src_agg_bezier_arc.cpp.o: plugin needed to
> handle lto object
>
> So it's probably ar which needs to be replaced by gnu-ar, don't know
> how, though:
> https://gcc.gnu.org/legacy-ml/gcc-help/2016-01/msg00061.html
> If you can manage this, and rebuild, fine. If not,
> can you try merging the branch pull/3/head of
> https://github.com/cxzhong/sage pull/3/head and see if it builds.
> This is an update of matplotlib and its deps, hopefully it will fix this all.

This should be all here now:
https://github.com/sagemath/sage/pull/40633

[Philippe Delavalade]

Hi Dima and many thanks for your help.

Le jeudi 21 août à 23:42, Dima Pasechnik a écrit :

> On Thu, Aug 21, 2025 at 1:19 PM Dima Pasechnik <dim...@gmail.com> wrote:
> >
> > Thanks for the log; one place there looks suspicious, and probably
> > results in your errors
> >
> > [spkg-install] [9/30] Linking static target extern/agg24-svn/libagg.a
> > [spkg-install] ar:
> > extern/agg24-svn/libagg.a.p/src_agg_bezier_arc.cpp.o: plugin needed to
> > handle lto object
> >
> > So it's probably ar which needs to be replaced by gnu-ar, don't know
> > how, though:

On my distro, its gnu-ar from binutils-2.37 ; do you think it's too old ?

> > https://gcc.gnu.org/legacy-ml/gcc-help/2016-01/msg00061.html
> > If you can manage this, and rebuild, fine. If not,
> > can you try merging the branch pull/3/head of
> > https://github.com/cxzhong/sage pull/3/head and see if it builds.
> > This is an update of matplotlib and its deps, hopefully it will fix this all.
>
> This should be all here now:
> https://github.com/sagemath/sage/pull/40633

As I've never used git, I aks some help to a friend of mine which the
maintener of slint.

I follow the commands he gave me :

git clone -c core.symlinks=true --filter blob:none \
--origin upstream --branch develop --tags \
https://github.com/sagemath/sage.git
cd sage
git fetch upstream pull/40633/head:pr-40633
git merge pr-40633
make configure
export SAGE_ROOT=~/sage/
MAKE="make -j8" # instaid of MAKEFLAGS="-j$(nproc) -l$(nproc).5"
# with this variable compilation fails for m4ri
./configure --disable-notebook \
--disable-doc \
--disable-r \
--with-system-python3=no
make

Compilation begins but after saying
[linbox-1.7.0+sage-2024-05-18] successfully installed
nothing happens ; After five or or ten minutes, I stopped.

Two tries gave the same thing.

Have you some idea or my commands are not good ?

Cheers.

--
Philippe

> To view this discussion visit https://groups.google.com/d/msgid/sage-support/CAAWYfq2uEKbZy0f0V1dMyEeJmhc1fOph%2BRXefVkQ%3DK%2BwCVNrTg%40mail.gmail.com.

[Dima Pasechnik]

On August 24, 2025 9:48:30 AM CDT, Philippe Delavalade <philippe....@gmail.com> wrote:
>Hi Dima and many thanks for your help.
>
>Le jeudi 21 août à 23:42, Dima Pasechnik a écrit :
>> On Thu, Aug 21, 2025 at 1:19 PM Dima Pasechnik <dim...@gmail.com> wrote:
>> >
>> > Thanks for the log; one place there looks suspicious, and probably
>> > results in your errors
>> >
>> > [spkg-install] [9/30] Linking static target extern/agg24-svn/libagg.a
>> > [spkg-install] ar:
>> > extern/agg24-svn/libagg.a.p/src_agg_bezier_arc.cpp.o: plugin needed to
>> > handle lto object
>> >
>> > So it's probably ar which needs to be replaced by gnu-ar, don't know
>> > how, though:
>
>On my distro, its gnu-ar from binutils-2.37 ; do you think it's too old ?

2.37 has been released 4 years ago. Up to date distros are on 2.42-2.45, it seems.

Do you have an reasonably new system Python3 (i.e., why do you use with-system-python3=no)?

I recommend to not do anything with MAKE and SAGE_ROOT.

Some versions of make are a bit buggy.