error building sage-9.3 from source

33 views
Skip to first unread message

Greg Marks

unread,
May 19, 2021, 4:53:48 AM5/19/21
to sage-s...@googlegroups.com
On a Debian 10 system, building sage-9.3 from source fails with this
error:

...
[dochtml] Traceback (most recent call last):
[dochtml] File "/usr/lib/python3.7/runpy.py", line 183, in _run_module_as_main
[dochtml] mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
[dochtml] File "/usr/lib/python3.7/runpy.py", line 142, in _get_module_details
[dochtml] return _get_module_details(pkg_main_name, error)
[dochtml] File "/usr/lib/python3.7/runpy.py", line 109, in _get_module_details
[dochtml] __import__(pkg_name)
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage_docbuild/__init__.py", line 57, in <module>
[dochtml] import sage.all
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/all.py", line 142, in <module>
[dochtml] from sage.rings.all import *
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/all.py", line 50, in <module>
[dochtml] from .finite_rings.all import *
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/finite_rings/all.py", line 21, in <module>
[dochtml] from .finite_field_constructor import FiniteField
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/finite_rings/finite_field_constructor.py", line 182, in <module>
[dochtml] from .finite_field_givaro import FiniteField_givaro
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/finite_rings/finite_field_givaro.py", line 24, in <module>
[dochtml] from sage.rings.finite_rings.element_givaro import Cache_givaro
[dochtml] File "sage/rings/finite_rings/element_givaro.pyx", line 1, in init sage.rings.finite_rings.element_givaro (build/cythonized/sage/rings/finite_rings/element_givaro.cpp:20735)
[dochtml] File "sage/rings/finite_rings/element_pari_ffelt.pyx", line 35, in init sage.rings.finite_rings.element_pari_ffelt (build/cythonized/sage/rings/finite_rings/element_pari_ffelt.c:14998)
[dochtml] File "sage/rings/polynomial/polynomial_element.pyx", line 67, in init sage.rings.polynomial.polynomial_element (build/cythonized/sage/rings/polynomial/polynomial_element.c:106651)
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/polynomial/polynomial_ring.py", line 154, in <module>
[dochtml] import sage.rings.polynomial.polynomial_element_generic as polynomial_element_generic
[dochtml] File "/usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/polynomial/polynomial_element_generic.py", line 1533, in <module>
[dochtml] from sage.rings.polynomial.polynomial_rational_flint import Polynomial_rational_flint
[dochtml] ImportError: /usr/local/Math_Software/sage-9.3/local/lib/python3.7/site-packages/sage/rings/polynomial/polynomial_rational_flint.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _Z27_arb_fmpz_poly_evaluate_arbP10arb_structPKllPKS_l
make[3]: *** [Makefile:2280: doc-html] Error 1
make[2]: *** [Makefile:2160: all-start] Error 2
make[2]: Leaving directory '/usr/local/Math_Software/sage-9.3/build/make'

I've tried it both with "./configure --with-system-flint=no" and with
"./configure --with-system-arb=no" and with a simple "./configure";
same problem each way.

Best regards,
Greg Marks
signature.asc

Greg Marks

unread,
May 19, 2021, 6:37:42 PM5/19/21
to sage-s...@googlegroups.com
For the record, this problem was solved by starting the build with
"./configure --with-system-python3=no"; then "make" completes without
error and "make ptestlong" shows all sage-9.3 tests passed.

Best regards,
Greg Marks

P.S. On some machines with Python 3.7.3 installed, the problem described
above arises; on others it doesn't. I suspect the problem has something
to do with the specific Python packages installed and is possibly
related to the (currently unanswered) question posed here:

https://stackoverflow.com/questions/67117097/c-cpython-38-x86-64-linux-gnu-so-undefined-symbol-zn6caffe28typemeta21-typem
signature.asc

Matthias Koeppe

unread,
May 19, 2021, 6:56:13 PM5/19/21
to sage-support
We'd need to see the top-level config.log corresponding to a failing build.

julian...@gmail.com

unread,
May 19, 2021, 7:26:45 PM5/19/21
to sage-support
Hi Greg,

Could you check what `libflint-arb.so*` files and `libarb.so*` files are on your system? I suspect that an old libarb is being picked up somehow.

On Thursday, May 20, 2021 at 12:37:42 AM UTC+2 Greg Marks wrote:
undefined symbol: _Z27_arb_fmpz_poly_evaluate_arbP10arb_structPKllPKS_l
This symbol has been present in arb since 2.12. Debian stretch has 2.8 so if somehow a libarb.so from the system gets picked up there it would explain it.
 
[...] related to the (currently unanswered) question posed here:

https://stackoverflow.com/questions/67117097/c-cpython-38-x86-64-linux-gnu-so-undefined-symbol-zn6caffe28typemeta21-typem

I don't think it's related btw.

julian
Reply all
Reply to author
Forward
0 new messages