fix build #27836
Closed
fix build #27836
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
ENH: Add Cython enumeration for NPY_FR_GENERIC
While the upper version is technically correct for the released version of NumPy (we are sure it will not work on Python 3.13) advertising it creates some problems, mostly for locking resolvers. They try to guess correct versions for non-released Python versions... This is probably an ecosystem or just "reasons", but it seems less useful than trouble to do the correct advertising here. See numpygh-24810 for *way* too much discussion about the why (and you will still be confused afterwards probably, so...). This needs to be fixed or at least documented clearer upstream by PyPA or similar, but... Closes numpygh-24810
REL: Remove Python upper version from the release branch
When a program located by find_program is executable directly, meson will execute it. If it is a script, its shebang (#!) line will control which interpreter is used. tempita.py specifies a standard shebang line of #!/usr/bin/env python3, so that python3 can be located by PATH lookup. In cases where python3 as found in the PATH is not correct, such as on systems that have multiple copies of python3 installed when building numpy for one that is not the python3 found by searching PATH, this interpreter will be wrong. When the python3 found by this mechanism doesn't have Cython available, it will cause the numpy build to fail. With the executable bit removed from tempita.py, meson will scan the script when attempting to execute it, locating the shebang line and substituting the appropriate sys.executable. See https://github.com/numpy/meson/blob/22df45a31981874310a78dde0df59a6a7c5ebb29/mesonbuild/programs.py#L179.
Replacing the original error is just not useful and actively unhelpful since the original array may have more information. We could chain the error, but there seems little reason to do so.
BLD: Use the correct Python interpreter when running tempita.py
MAINT: Remove unhelpful error replacements from ``import_array()``
This is a custom-built wheel that can only be detected through pkg-config. So trying other methods just costs time and increases the size of `meson-log.txt`.
BLD: use classic linker on macOS, the new one in XCode 15 has issues
BLD: musllinux_aarch64 [wheel build]
[skip ci]
MAINT: Update mailmap
This is a backport of numpy#24493 and numpy#24291. The purpose of this is to ease future backports that expect these files. - CI: move some jobs in `build_test.yml` to Meson - CI: split `build_test.yml` into three GHA jobs files Also documents better what is being run. See numpygh-24410 for the overall restructuring plan for GitHub Actions CI. - CI: merge `linux_meson.yml` into `linux_blas.yml` - TST: disable mypy tests in test suite unless an environment variable is set These tests are super slow, and they're effectively always passing in CI. Running them on all "full" test suite runs is too expensive. Note that SciPy has an XSLOW mark, NumPy does not. So use an env var for now. - CI: add new GHA CI jobs to run MyPy across OS/Python flavors
typing_nptensions -> typing_extensions
Backport of numpy#24625.
MAINT: Split up .github/workflows to match main
Backports of numpy#24969, numpy#24979, numpy#24968, numpy#25068. * apply 24969.diff * apply 24979.diff * apply 24968.diff * TST: skip flaky test in test_histogram --------- Co-authored-by: mattip <matti.picus@gmail.com>
This fix addresses two issues:
* Corrects the use of unsupported instructions by the assembler
in half-precision to double-precision conversion.
* Resolves a code error related to variable naming during conversion.
It also:
* Implement a compile-time test on PPC64 to determine support for Half/Double
VSX3 instructions by the assembler
* Modify half-precision tests to deal with FP HW exceptions
MAINT: align test_dispatcher s390x targets with _umath_tests_mtargets
BUG: Backport fix build on ppc64 when the baseline set to Power9 or higher
Fixes function blocklisting for glibc<2.18, reported in issue numpygh-25087. Signed-off-by: Marcel Bargull <marcel.bargull@udo.edu>
This is the minimal backport version of numpygh-25094, which simply aligns the Cython 2 and Cython 3 definitions.
Signed-off-by: mattip <matti.picus@gmail.com>
BUG: Avoid intp conversion regression in Cython 3 (backport)
TST: Fix test_numeric on riscv64
Using a local prototype for fallocate instead of the fcntl.h header meant that the redirect triggered by -D_FILE_OFFSET_BITS=64 was not triggered. The prototypes in feature_detection_stdio.h should only be used by functions in setup_common.py. If they are used by the feature discovery code they might trigger false positives.
It currently raises a warning: ``` DeprecationWarning: Attribute s is deprecated and will be removed in Python 3.14; use value instead ``` AFAICT, the `value` attribute has always existed (at least in supported versions of Python 3.)
This version fixes the DeprecationWarning in Python 3.12: ``` DeprecationWarning: datetime.datetime.utcfromtimestamp() is deprecated and scheduled for removal in a future version. Use timezone-aware objects to represent datetimes in UTC: datetime.datetime.fromtimestamp(timestamp, datetime.UTC). ```
TST: Fix test_warning_calls on Python 3.12
TST: Bump pytz to 2023.3.post1
BUG: Fix AVX512 build flags on Intel Classic Compiler
Backport of numpy#25669. The current code isn't robust to backslashes (from path segments on Windows) in linker flags. No bug report for NumPy, but the same problem did show up in SciPy already. Fix is the same as scipy/scipy#19945. This actually pulls in all of `__config__.py` from main because there have been a number of other changes. [skip cirrus] [skip circle]
BLD: fix potential issue with escape sequences in ``__config__.py``
BUG: Use large file fallocate on 32 bit linux platforms
CI: pin cygwin python to 3.9.16-1 and fix typing tests [skip cirrus][skip azp][skip circle]
MAINT: Bump cibuildwheel to v2.16.4
(cherry picked from commit 9b38c63)
The meson-python 0.15.0 release added the ability to configure a custom `meson` in `pyproject.toml`. So there's no longer a need for us to vendor it. (cherry picked from commit 6544e33, adding a version range to depend on meson-python 0.15.x)
This fixes issues with Windows wheel builds; a new GitHub Actions runner image caused argument escaping to break.
BLD: unvendor meson-python on 1.26.x and upgrade to meson-python 0.15
* BUG: Fix np.quantile([Fraction(2,1)], 0.5) * address review comments * pass type instead of dtype instance
MAINT: Include header defining backtrace
BUG: Fix np.quantile([Fraction(2,1)], 0.5) (numpy#24711)
BUG: Restore missing asstr import
- Create 1.26.4-changelog.rst - Update 1.26.4-notes.rst - Remove old fragments from doc/release/upcoming_changes [wheel build]
REL: Prepare for the NumPy 1.26.4 release
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
28 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
NPY_TARGET_VERSIONrelease notedownload-wheels.np.dtypemetadata parametermetadataparameter of thedtypeconstructor--allow-downgradeoption for rtools.return NULLto begoto failprivateentities #23879 for private proceduresprivateentities #23879 on private module proceedures-ftrapping-mathwith Clang on macOSmeson_options.txtnp.__all__getenvcall used for memory policy warningtools/wheels/for the move to Mesonrun_commandcall_NestedSequence.__getitem__signaturememcpyand swap orderextbuild.pyfrom main.pipon building docs with spinasv devhas been removed, useasv run.typing.ConcatenateMYPYPATHtyping_extensionsusage for <py310, add it back in environment.yml_umath_linalgdependencies (BLD: fix_umath_linalgdependencies #24584)castingkeyword tonp.clipfromnumeric.pyif2pydoesn't mapiso_c_bindingvalues by default #24553binary_reprto accept any object implementing__index__dtypeandgenerichashabletyping.assert_typetyping.assert_typenp.ctypeslibtest fixes for windowsmesonbackend forf2py(ENH:mesonbackend forf2py#24532)collections.abc.Bufferin the Array APIcollections.abc.Bufferin thenpt.ArrayLikedefinitionspin docscommand-march=nativeon unsupported architectures-fimplicit-none#24550use-compute-creditsfor Cirrus.spin/cmds.pyfrom main.linux_musl.ymlfrom main.pyproject.toml_distributor_init.pychange to support scipy-openblasimport_array()test_*installed)noexceptto shuffle helpersallow-noblasoption to true.np.dtypeto itself doesn't crash__getitem__innumpy.array_api(TYP: add None to__getitem__innumpy.array_api#25022)newaxisandlinalg.solveinnumpy.array_apilongtypebasein cpu_avx512_knnumpy/f2py/_backendsfrom main.f2py/*.pyfrom main.newaxisto__all__innumpy.array_api__config__.pydistutilsin_core/code_generators/