Update conda-lock files, rename macOS conda-lock files to match CI Conda, remove unexplained duplicate lock files

[mkoeppe]

The renaming is done so that the fix here can take effect on PR runs via the "CI Fix" mechanism (as changes to the workflow files themselves cannot take effect).

• Moreover, making arm64 the default platform for macOS makes sense because all new macOS machines use this platform.

And update by running .github/workflows/conda-lock-update.py for all architectures.

• Workflow matrix variables now contain -latest, to add tests for the macOS x86_64 platform (runner name: macos-13)

• Removes some unexplained duplications (both arm and arm64 files for macOS, both aarch and aarch64 for Linux)

• Fixes CI Conda is broken because macos-latest is now Apple Silicon #37997

Author: @mkoeppe, @gmou3

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies

• Depends on Avoid conflict with macro minor #38033 (split out from here)
• Depends on sage.libs.giac: Compile with std=c++11 #38034 (split out from here)
• Depends on Reject system gap 4.13 #38222 (merged here)

[github-actions]

Documentation preview for this PR (built with commit b560f4d; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[mkoeppe]

  File "sage/algebras/quatalg/quaternion_algebra_cython.pyx", line 1, in init sage.algebras.quatalg.quaternion_algebra_cython
  File "sage/matrix/matrix_integer_dense.pyx", line 1, in init sage.matrix.matrix_integer_dense
ImportError: dlopen(/Users/runner/work/sage/sage/src/sage/libs/linbox/linbox_flint_interface.cpython-311-darwin.so, 0x0002): symbol not found in flat namespace '__ZN6FFPACK8CharPolyIN6Givaro8Poly1DomINS1_7ModularIddvEENS1_5DenseEEEEERNT_7ElementERKS7_S9_mNS7_8Domain_t11Element_ptrEmRNSC_8RandIterENS_19FFPACK_CHARPOLY_TAGEm'

[mkoeppe]

  In file included from /Users/runner/miniconda3/envs/sage/include/giac/giac.h:5:
  In file included from /Users/runner/miniconda3/envs/sage/include/giac/poly.h:26:
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:341:10: error: no template named 'pointer_to_binary_function' in namespace 'std'
      std::pointer_to_binary_function < const monomial<T> &, const monomial<T> &, bool> strictly_greater ;
      ~~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:342:28: error: no template named 'pointer_to_binary_function' in namespace 'std'
      sort_helper(const std::pointer_to_binary_function < const monomial<T> &, const monomial<T> &, bool> is_strictly_greater):strictly_greater(is_strictly_greater) {};
                        ~~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:343:41: error: no member named 'ptr_fun' in namespace 'std'
      sort_helper():strictly_greater(std::ptr_fun<const monomial<T> &, const monomial<T> &, bool>(m_lex_is_strictly_greater<T>)) {};
                                     ~~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:343:49: error: expected expression
      sort_helper():strictly_greater(std::ptr_fun<const monomial<T> &, const monomial<T> &, bool>(m_lex_is_strictly_greater<T>)) {};
                                                  ^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:343:70: error: expected expression
      sort_helper():strictly_greater(std::ptr_fun<const monomial<T> &, const monomial<T> &, bool>(m_lex_is_strictly_greater<T>)) {};
                                                                       ^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:343:95: error: expected '(' for function-style cast or type construction
      sort_helper():strictly_greater(std::ptr_fun<const monomial<T> &, const monomial<T> &, bool>(m_lex_is_strictly_greater<T>)) {};
                                                                                            ~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:680:18: error: no template named 'pointer_to_binary_function' in namespace 'std'
               const std::pointer_to_binary_function < const monomial<T> &, const monomial<T> &, bool> m_is_strictly_greater
                     ~~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:729:44: error: no template named 'pointer_to_binary_function' in namespace 'std'
      typedef std::map< index_t,T,const std::pointer_to_binary_function < const index_m &, const index_m &, bool> > application;
                                        ~~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:730:30: error: no type named 'ptr_fun' in namespace 'std'
      application produit(std::ptr_fun(is_strictly_greater));
                          ~~~~~^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:737:14: error: 'application' (aka 'int') is not a class, namespace, or enumeration
      typename application::iterator prod_it,prod_itend;
               ^
  /Users/runner/miniconda3/envs/sage/include/giac/monomial.h:754:29: error: use of undeclared identifier 'prod_itend'; did you mean 'prod_it'?
      prod_it=produit.begin(),prod_itend=produit.end();

[mkoeppe]

That's with giac 1.9.0.21 h1c96721_1 conda-forge

[mkoeppe]

https://en.cppreference.com/w/cpp/utility/functional/pointer_to_binary_function
(removed in C++17)

[mkoeppe]

  building 'sage.matroids.graphic_matroid' extension
  /home/runner/miniconda3/envs/sage/bin/x86_64-conda-linux-gnu-cc -Wno-unused-result -Wsign-compare -DNDEBUG -fwrapv -O2 -Wall -fPIC -O2 -isystem /home/runner/miniconda3/envs/sage/include -fPIC -O2 -isystem /home/runner/miniconda3/envs/sage/include -march=nocona -mtune=haswell -ftree-vectorize -fPIC -fstack-protector-strong -fno-plt -O2 -ffunction-sections -pipe -isystem /home/runner/miniconda3/envs/sage/include -DNDEBUG -D_FORTIFY_SOURCE=2 -O2 -isystem /home/runner/miniconda3/envs/sage/include -fPIC -Isage/data_structures -I/home/runner/miniconda3/envs/sage/lib/python3.9/site-packages/cysignals -Isage/cpython -I/home/runner/work/sage/sage/src -I/home/runner/miniconda3/envs/sage/lib/python3.9/site-packages/numpy/core/include -I/home/runner/miniconda3/envs/sage/include/python3.9 -I/home/runner/miniconda3/envs/sage/include/python3.9 -c sage/matroids/graphic_matroid.c -o /tmp/tmpj1q6x7la.build-temp/sage/matroids/graphic_matroid.o
  sage/matroids/graphic_matroid.c: In function '__pyx_f_4sage_8matroids_15graphic_matroid_14GraphicMatroid__has_minor':
  sage/matroids/graphic_matroid.c:23551:234: error: macro "minor" passed 3 arguments, but takes just 1
  23551 |       __pyx_t_2 = ((struct __pyx_vtabstruct_4sage_8matroids_15graphic_matroid_GraphicMatroid *)__pyx_v_self->__pyx_base.__pyx_vtab)->__pyx_base.minor(((struct __pyx_obj_4sage_8matroids_7matroid_Matroid *)__pyx_v_self), 0, &__pyx_t_21); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 610, __pyx_L1_error)
        |                                                                                                                                                                                                                                          ^
  In file included from /home/runner/miniconda3/envs/sage/x86_64-conda-linux-gnu/sysroot/usr/include/sys/types.h:226,
                   from /home/runner/miniconda3/envs/sage/x86_64-conda-linux-gnu/sysroot/usr/include/stdlib.h:320,
                   from /home/runner/miniconda3/envs/sage/include/python3.9/Python.h:34,
                   from sage/matroids/graphic_matroid.c:50:
  /home/runner/miniconda3/envs/sage/x86_64-conda-linux-gnu/sysroot/usr/include/sys/sysmacros.h:66: note: macro "minor" defined here
     66 | # define minor(dev) gnu_dev_minor (dev)
        |

[mkoeppe]

@gmou3 Here we have run in a strange symbol clash from the Cythonization of GraphicMatroid. See https://stackoverflow.com/questions/22240973/major-and-minor-macros-defined-in-sys-sysmacros-h-pulled-in-by-iterator; would you be able to work around it?

[vbraun]

Shouldn't the CI then work on this ticket if this ticket fixes it? Because it seems like @mkoeppe canceled the run. Should it be this way?

[mkoeppe]

@vbraun This fixes the "Build&Test using conda": See the green checkmarks.

[mkoeppe]

@vbraun I've now updated the branch protection rule to make the "Expected -- Waiting for status to be reported" disappear (for conda)

[mkoeppe]

Thanks, Volker.

[mkoeppe]

@vbraun I've redone the branch so that the conda package update from #35396 is reflected.

[mkoeppe]

Tests at https://github.com/mkoeppe/sage/actions/runs/9671105316

[mkoeppe]

Tests at https://github.com/mkoeppe/sage/actions/runs/9672092917

[kwankyu]

• Test run before this commit: https://github.com/sagemath/sage/actions/runs/9525058035/job/26258729278
• Test run after this commit: https://github.com/sagemath/sage/actions/runs/9525536510/job/26259809272
  Same errors can also be found in Update conda-lock files, rename macOS conda-lock files to match CI Conda, remove unexplained duplicate lock files #37998 (comment)

False facts. What Tobias cites as a "test run before this commit" is actually a test run with an earlier version of this PR merged as a CI Fix.

Confirmed. "Test run before this commit" has the same failures but in the baseline set from this PR merged as a CI Fix.

Tobias' argument (no upgrade that introduces new failures) is sound but based on false facts.

[kwankyu]

Is this

sage -t --random-seed=147712982467446710824553287018105912799 src/sage/libs/singular/function.pyx  # 3 doctests failed

new?

[kwankyu]

As we are here, may I ask (not a thing to do here): Could we put all these environment-xxx files into, say conda directory to clean up the sage root directory, and recommend

conda env create --file ./conda/environment-3.11-linux.yml --name sage-build

in the developer guide?

[mkoeppe]

some restructuring like this could be a good idea, but not for this PR

[mkoeppe]

Is this

sage -t --random-seed=147712982467446710824553287018105912799 src/sage/libs/singular/function.pyx  # 3 doctests failed

new?

I think this is another sporadic failure, which I've probably seen before.
It does not show in the set of tests https://github.com/mkoeppe/sage/actions/runs/9672092917

[kwankyu]

Another question. This is also not to be done here:

It seems that the script .github/workflows/conda-lock-update.py is a utility for humans. Why is it in .github/workflows?

[mkoeppe]

It was said at the time that maybe one day these lock files would be updated automatically by a GH Actions workflow.

[kwankyu]

Regarding

The renaming of the macos lock files are now inconsistent with the naming of the linux lock files. We should not orient ourselves on what github actions use as the default, but what is the most common system - these files are primarily intended to be used by users, and not for the ci.

As far as I can see, there is nothing "oriented by github actions" in naming lock files. But the comment may be to the old version of this PR...

[kwankyu]

It was said at the time that maybe one day these lock files would be updated automatically by a GH Actions workflow.

Yes, I thought something like it. Perhaps ci-conda.yml may run the script and deliver the diffs of updated lock files as an artifact or as annotations...

[kwankyu]

I am positive to this PR. We need this PR, ASAP to fix conda ci.

[mkoeppe]

Thanks.