It is a bit too obscure, but explaining that it happens for datetime
unit conversions seems more...

This PR Closes gh-17046.

The problem was that when calling mask=None, the array creation took...

Co-authored-by: Matti Picus <[email protected]>

BENCH: Add a test for masked array creations

BUG: Fix some valgrind errors (and probably harmless warnings)

* initial files

* STY: remove noop.txt (we have an environment.yaml)

some other style fixu...

TST: Skip when numba/numpy compat issues cause SystemError

numba is a bit buggy when wrapping ufuncs, so what should be nothing
is (on non dev versions) a ...

DOC: misleading text lead to false hope

BLD: fix cirrus wheel upload triggers

BUG: Fix <complex 0>^{non-zero}

Co-authored-by: Marten van Kerkwijk <[email protected]>

It appears that assertion is not true for bad __array_ufunc__
or similar implementations. And w...

In theory, an object matmul could be caught here, but lets assume
that doesn't happen...

Add special casing for `1d @ 1d` and `2d @ 1d` ops.

In principle, this is probably still not 100% correct always since
we convert the the other obje...

This uses the `axes` argument. Arguably the most correct version since
we use the full ufunc ma...

Python does even crash when you get it wrong, so this is unnecessar

DOC: Remove dangling deprecation warning

just silences a compiler warning.

This deprecation is no longer mentioned elsewhere on the page.

Co-authored-by: Matti Picus <[email protected]>

Co-authored-by: Matti Picus <[email protected]>

Co-authored-by: Matti Picus <[email protected]>

This allows correctly cathing the error, also adjust the the NoLoop
error is used for the the "b...

This finalize the deprecation warning, there are a couple of corner
cases that we don't get as n...

If promotion fails internally, this effectively means that there is
no loop available for the gi...

MAINT: unify NPY_NO_SIGNAL macros

[ci skip]

API: Add new `np.exceptions` namespace for errors and warnings

BLD: revert adding PEP 621 metadata, it confuses setuptools

CI: Make benchmark asv run quick to only check that benchmarks work

This means that benchmark results are effectively useless but I do not
think we use them anyway ...

Co-authored-by: Matti Picus <[email protected]>

Both should now live in the "exceptions" module.

This means moving ComplexWarning, TooHardError, and AxisError.

AxisError did exist, but e.g. ComplexWarning wasn't even properly
included.

MAINT: Add `np._utils` to meson

Odd PR timing/based on an older branch or seems to have hid,
this necessary addition when adding...

MAINT: Move set_module from numpy.core to numpy._utils

API: Add numpy.testing.overrides to aid testing of custom array containers

See comments on gh-22663

NumPy already has SSE2 versions of `negative`. Changes here convert that to universal intrinsic...

API: (cython) remove `long_t` and `ulong_t`

BUILD: move wheel uploads to cirrus for aarch64

* DOC: Slightly improve error when gufunc axes has wrong size

My hope was to tweak it into so...

* make isnan, isinf, isfinite, signbit, nextafter aliases

* fixes from review

Co-authored-...

CI: Add cirrus-ci to test linux_aarch64

BUG: Polynomials now copy properly (#22669)

BUG: Ensure string aliases `"int0"`, etc. remain valid for now

Co-authored-by: Matti Picus <[email protected]>

BLD: revert function() -> #define for 3 npymath functions

MAINT: fix typo in loops_unary_fp.dispatch.c.src

MAINT: fix `.gitignore` issues, and remove remaining `NPY_INLINE` usages