[MementoRC]

1.6.0 added EXTENSION_SUFFIXES to importlib-machinery-patch so that setuptools get_abi3_suffix() stops picking the build interpreter's platform-tagged .abi3-<multiarch>.so on a cross build. The second element is hardcoded to ".abi3.so", which is correct only for a GIL-enabled host.

In Python/dynload_shlib.c the plain abi3 entries are inside the #ifndef Py_GIL_DISABLED block while the abi3t entries are outside it, so a free-threaded interpreter has no plain .abi3 entry in its filetab at all. With the hardcoded list, a free-threaded cross build names the extension .abi3.so, which that interpreter cannot load.

This takes the suffix from the host's own Py_GIL_DISABLED instead. .get() rather than indexing, since the var is absent from a GIL build's build_time_vars rather than set to 0.

Measured on a GIL build, CPython 3.15.0rc2 x86_64, _imp.extension_suffixes():

['.cpython-315-x86_64-linux-gnu.so', '.abi3-x86_64-linux-gnu.so', '.abi3.so',
 '.abi3t-x86_64-linux-gnu.so', '.abi3t.so', '.so']

I have not tested a free-threaded cross build; this is from reading dynload_shlib.c and the host build_time_vars. Happy to adjust if you would rather list both forms than pick one - note that get_abi3_suffix() takes the first entry containing .abi3, and .abi3t.so matches that substring, so ordering would become load-bearing.

Raised by @isuruf in conda-forge/crossenv-feedstock#49.

[isuruf]

This is only for 3.15+. For earlier versions there's no abi3.

[MementoRC]

Ok, so this calls for a version guard, a dedicated helper function?

[isuruf]

We are also missing abi3-x86_64-linux-gnu.so types coming in 3.15

[MementoRC]

So, abi-<arch>.so. Could try extracting from sysconfig_ext_suffix

[MementoRC]

Is the order correct?

        suffixes = [".%s.so" % stem]
        soabi_platform = self._host_soabi_platform()
        if soabi_platform:
            suffixes.append(".%s-%s.so" % (stem, soabi_platform))

[MementoRC]

New test is passing: tests/test_environment.py::test_abi3_extension_suffixes PASSED

[auscompgeek]

My reading of https://docs.python.org/3.15/howto/abi3t-migration.html is that abi3t is compatible with both freethreaded and non-freethreaded CPythons starting from 3.15. I don't believe this correctly implements that?

I can see rules_pycross appears to handle it correctly.

[MementoRC]

Right, .abi3t is unconditional