This is an ABI incompatibility with clang 14 and 15 where givaro was compiled with 14 and sage was compiled with 15. I suggest downgrading to clang 14 or set flags like https://github.com/conda-forge/sagelib-feedstock/blob/5542b278f642b772ae243b323c7cbc7a170a5f9a/recipe/build.sh#L21-L25 until we rebuild givaro with clang 15.
Isuru On Wed, May 31, 2023 at 2:01 PM Matthias Goerner <enisc...@gmail.com> wrote: > I am focusing on the Sage 9.8 issue of the missing symbol > __ZN6FFPACK8CharPolyIN6Givaro8Poly1DomINS1_7ModularIddvEENS1_5DenseEEEEERNT_7ElementERKS7_S9_mNS7_8Domain_t11Element_ptrEmRNSC_8RandIterENS_19FFPACK_CHARPOLY_TAGEm > > Demangled with c++filt, the missing symbol is: > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Element& > FFPACK::CharPoly<Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense> >(Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense> const&, Givaro::Poly1Dom<Givaro::Modular<double, double, > void>, Givaro::Dense>::Element&, unsigned long, > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Domain_t::Element_ptr, unsigned long, > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Domain_t::RandIter&, FFPACK::FFPACK_CHARPOLY_TAG, unsigned > long) > > I expect this to be in ffpack, so I made sure that ffpack is indeed picked > up from the dedicated conda environment: > > lsof -p 70202 | grep -i ffpack > python3.1 70202 matthias txt REG 1,15 1132624 5917685 > /Users/matthias/anaconda3/envs/sage98-py311/lib/libffpack.1.dylib > > I then looked whether similar symbols are available in ffpack with the > following command: > nm /Users/matthias/anaconda3/envs/sage98-py311/lib/libffpack.1.dylib | > grep > __ZN6FFPACK8CharPolyIN6Givaro8Poly1DomINS1_7ModularIddvEENS1_5DenseEEEEERNT_7ElementERKS7_S9_mNS7_8Domain_t11Element_ptrEm > > Which gave: > 000000000004f17c T > __ZN6FFPACK8CharPolyIN6Givaro8Poly1DomINS1_7ModularIddvEENS1_5DenseEEEEERNT_7ElementERKS7_S9_mNS7_8Domain_t11Element_ptrEmNS_19FFPACK_CHARPOLY_TAGEm > 000000000004ec10 T > __ZN6FFPACK8CharPolyIN6Givaro8Poly1DomINS1_7ModularIddvEENS1_5DenseEEEEERNT_7ElementERKS7_S9_mNS7_8Domain_t11Element_ptrEmRNS7_8Domain_t8RandIterENS_19FFPACK_CHARPOLY_TAGEm > > which demangled gives: > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Element& > FFPACK::CharPoly<Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense> >(Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense> const&, Givaro::Poly1Dom<Givaro::Modular<double, double, > void>, Givaro::Dense>::Element&, unsigned long, > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Domain_t::Element_ptr, unsigned long, > FFPACK::FFPACK_CHARPOLY_TAG, unsigned long) > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Element& > FFPACK::CharPoly<Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense> >(Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense> const&, Givaro::Poly1Dom<Givaro::Modular<double, double, > void>, Givaro::Dense>::Element&, unsigned long, > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Domain_t::Element_ptr, unsigned long, > Givaro::Poly1Dom<Givaro::Modular<double, double, void>, > Givaro::Dense>::Domain_t::RandIter&, FFPACK::FFPACK_CHARPOLY_TAG, unsigned > long) > > And then I got confused: the latter demangled method has the same > signature as the missing one. > > In other words: from a C++ perspective, the missing method is there, it is > just that the way the name and signature get mangled is different between > linbox_flint_interface.cpython-311-darwin.so and libffpack.1.dylib > > So it seems it is an ABI problem where we get two different managled names > for the same function signature. > > I am not sure what to make out of this and have only two explanations: > 1. It is a bug in the C++ compiler. > 2. The two libraries were compiled with different C++ compilers or with > different compiler options. > > On Wednesday, May 31, 2023 at 11:01:20 AM UTC-7 Matthias Goerner wrote: > > Sorry for the late reply. Here are the details about my conda environment. > I am not exactly sure what other information is useful to debug this. I am > familiar with the linux tools, but am struggling with otool and vmmap to > figure out what is going on here and see whether I am picking up any > library from something that is not in the dedicated conda environment. > > On Friday, May 26, 2023 at 8:05:04 AM UTC-7 Matthias Koeppe wrote: > > config.log please > > On Wednesday, May 24, 2023 at 3:16:25 PM UTC-7 Matthias Goerner wrote: > > Hi! > > I can't get sage to work when installing it with conda on a new MacBook > Pro with M2 chip and Ventura 13.3. > > I tried > conda -n sage create sage=9.X python=3.Y > for (X,Y) = (5,10), (8,10), (8,11). > With Sage 9.5, I get the attached crash report. > With Sage 9.8, I get the following linking error below. > > Cheers, > Matthias > > (sage-py310) matthias@matthiass-air-2 ~ % sage > ┌────────────────────────────────────────────────────────────────────┐ > │ SageMath version 9.8, Release Date: 2023-02-11 │ > │ Using Python 3.10.11. Type "help()" for help. │ > └────────────────────────────────────────────────────────────────────┘ > ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓ > ┃ Warning: sage.all is not available; this is a limited REPL. ┃ > ┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛ > sage: import sage.all > --------------------------------------------------------------------------- > ImportError Traceback (most recent call last) > Cell In[1], line 1 > ----> 1 import sage.all > > File > ~/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/all.py:168 > 166 from sage.modules.all import * > 167 from sage.monoids.all import * > --> 168 from sage.algebras.all import * > 169 from sage.modular.all import * > 170 from sage.sat.all import * > > File > ~/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/algebras/all.py:21 > 5 > #***************************************************************************** > 6 # Copyright (C) 2005 William Stein [...] > 7 # > (...) > 17 # http://www.gnu.org/licenses/ > 18 > #***************************************************************************** > 19 from sage.misc.lazy_import import lazy_import > ---> 21 import sage.algebras.catalog as algebras > 23 from .quatalg.all import * > 24 from .steenrod.all import * > > File > ~/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/algebras/catalog.py:84 > 1 r""" > 2 Catalog of Algebras > 3 > (...) > 80 <sage.algebras.yokonuma_hecke_algebra.YokonumaHeckeAlgebra>` > 81 """ > 83 from sage.algebras.free_algebra import FreeAlgebra as Free > ---> 84 from sage.algebras.quatalg.quaternion_algebra import > QuaternionAlgebra as Quaternion > 85 from sage.algebras.steenrod.steenrod_algebra import > SteenrodAlgebra as Steenrod > 86 from > sage.algebras.finite_dimensional_algebras.finite_dimensional_algebra import > FiniteDimensionalAlgebra as FiniteDimensional > > File > ~/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/algebras/quatalg/quaternion_algebra.py:68 > 61 from operator import itemgetter > 63 from .quaternion_algebra_element import ( > 64 QuaternionAlgebraElement_abstract, > 65 QuaternionAlgebraElement_generic, > 66 QuaternionAlgebraElement_rational_field, > 67 QuaternionAlgebraElement_number_field) > ---> 68 from . import quaternion_algebra_cython > 70 from sage.modular.modsym.p1list import P1List > 72 from sage.misc.cachefunc import cached_method > > File > ~/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/algebras/quatalg/quaternion_algebra_cython.pyx:1, > in init sage.algebras.quatalg.quaternion_algebra_cython > (build/cythonized/sage/algebras/quatalg/quaternion_algebra_cython.cpp:6484)() > ----> 1 # distutils: language = c++ > 2 # distutils: libraries = gmp m NTL_LIBRARIES > 3 # distutils: extra_compile_args = NTL_CFLAGS > > File > ~/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/matrix/matrix_integer_dense.pyx:1, > in init sage.matrix.matrix_integer_dense > (build/cythonized/sage/matrix/matrix_integer_dense.cpp:59322)() > ----> 1 # -*- coding: utf-8 -*- > 2 # distutils: extra_compile_args = NTL_CFLAGS M4RI_CFLAGS > 3 # distutils: libraries = iml NTL_LIBRARIES gmp m CBLAS_LIBRARIES > > ImportError: > dlopen(/Users/matthias/anaconda3/envs/sage-py310/lib/python3.10/site-packages/sage/libs/linbox/ > linbox_flint_interface.cpython-310-darwin.so, 0x0002): symbol not found > in flat namespace > '__ZN6FFPACK8CharPolyIN6Givaro8Poly1DomINS1_7ModularIddvEENS1_5DenseEEEEERNT_7ElementERKS7_S9_mNS7_8Domain_t11Element_ptrEmRNSC_8RandIterENS_19FFPACK_CHARPOLY_TAGEm' > > -- > 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+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/sage-support/c028a399-8ecc-4370-9130-88b303487e50n%40googlegroups.com > <https://groups.google.com/d/msgid/sage-support/c028a399-8ecc-4370-9130-88b303487e50n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- 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+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-support/CA%2B01voO9E7_w2qE5uYYT8AXRfQLbrpJfRGjk274HpJ0RyiMo5A%40mail.gmail.com.
