| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bumps [pytz](https://github.com/stub42/pytz) from 2021.1 to 2021.3.
- [Release notes](https://github.com/stub42/pytz/releases)
- [Commits](https://github.com/stub42/pytz/compare/release_2021.1...release_2021.3)
---
updated-dependencies:
- dependency-name: pytz
dependency-type: direct:production
update-type: version-update:semver-minor
...
Signed-off-by: dependabot[bot] <support@github.com>
|
|\
| |
| | |
MAINT: Test PyPy3.8
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
|\ \
| | |
| | | |
BUG: Fix the `lib.function_base` window functions ignoring extended precision float dtypes
|
| | | |
|
| | | |
|
| | |
| | |
| | |
| | | |
precision float dtypes
|
|\ \ \
| | | |
| | | | |
ENH: Add annotations for `np.lib.function_base` part 1
|
| | | |
| | | |
| | | |
| | | | |
`np.place` uses the `np.core.multiarray._insert` function under the hood, not to be confused with `np.insert` (which does accept slices)
|
| | | | |
|
| | | | |
|
| | | | |
|
| |/ / |
|
|\ \ \
| | | |
| | | | |
STY: Harmonize rules with cb_rules for f2py
|
| | | |
| | | |
| | | |
| | | | |
Co-authored-by: Sebastian Berg <sebastian@sipsolutions.net>
|
| | | | |
|
| | | | |
|
|\ \ \ \
| | | | |
| | | | | |
ENH: core: More informative error message for broadcast(*args)
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
When broadcast(*args) fails because of a shape mismatch, include
in the error message which arguments caused the mismatch and what
their shapes are.
For example, instead of
>>> np.broadcast([[0, 0, 0]], [[1], [1]], [2, 2])
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: shape mismatch: objects cannot be broadcast to a
single shape
we now get
>>> np.broadcast([[0, 0, 0]], [[1], [1]], [2, 2])
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ValueError: shape mismatch: objects cannot be broadcast to a
single shape. Mismatch is between arg 0 with shape (1, 3) and
arg 2 with shape (2,).
This also affects broadcast_arrays() and broadcast_shapes().
Closes gh-8345.
|
|\ \ \ \ \
| | | | | |
| | | | | | |
MAINT:redundant 'else' statement with 'for' loop#19077
|
| | | | | |
| | | | | |
| | | | | | |
Co-authored-by: Bas van Beek <43369155+BvB93@users.noreply.github.com>
|
| |/ / / / |
|
|\ \ \ \ \
| | | | | |
| | | | | | |
BUG: Add a warning for user dtypes modifying casts after use
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This adds a RuntimeWarning if a modification for a cast happens
after said cast was already used (which could happen accidentally).
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
* for `datetime64` units smaller than `s`, handle some
common/reasonable "metrix prefix" casting conversions like
1000 ns == 1 us
* this adds support specifically for equivalencies that span
10^3 and 10^6 fold, but there are even more beyond that, and if
we want to handle them all we may want to abstract to a handler
function
* that said, this still seems like a solid improvement that covers
some of the more common cases?
* MAINT: PR 19731 revisions
* add support/testing for datetime unit conversions at the
scale of 10^9, when below the 32-bit integer overflow limit
* raise a `TypeError` when `M8` unit magnitude exceeds
`INT_MAX`
* MAINT: PR 19731 revisions
* add new test `test_datetime_prefix_conversions()` that
performs bidirectional metric time prefix casting checks
* use `long long int` for `true_meta_num` on 32-bit systems
* some C code whitespace changes requested during review
* MAINT: PR 19731 revisions
* flake8-related fixes
* MAINT: PR 19731 revisions
* use a fixed size `int64_t` type for `true_meta_num`
on Windows to allow the overflow check to happen properly
* whitespace adjustments for the C code based on review comments
* MAINT: PR 19731 revisions
* improve the portability of the overflow
check for datetime unit integer prefixes
* MAINT: PR 19731 revisions
* simplify the changes by using `NumPyOS_strtoll` (less
custom C needed this way)
* MAINT: cleanup after merge conflict.
* MAINT: PR 19731 revisions
* based on reviewer feedback, make negative
datetime units a `TypeError` (including regression
test)
|
| |/ / / /
|/| | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
* BUG: Allow (arg-)partition to accept `uint64` indices
* DEP: Deprecate the use of booleans as (arg-)partition indices
* TST,DEP: Add tests for the `parametrize` bool-index-deprecation
* TST: Add more dtype-based tests for (arg-)partition
* DOC: Add a release note for the (arg-)partition boolean deprecation
* DEP: Explicitly mention numpy 1.22 in the (arg-)partition deprecation warning
Co-Authored-By: Sebastian Berg <sebastian@sipsolutions.net>
Co-authored-by: Sebastian Berg <sebastian@sipsolutions.net>
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
* DOC: add example showing how to convert POSIX timestamps to datetime64
Co-authored-by: Melissa Weber Mendonça <melissawm@gmail.com>
|
|\ \ \ \ \
| | | | | |
| | | | | | |
DOC: replace return type in np.ma.* docstring
|
| | | | | | |
|
| | | | | | |
|
| | | | | | |
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | | |
BUG: avoid infinite recurrence on dependencies in crackfortran
|
|/ / / / / / |
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This reverts commit 1ac3cbea412c7d9ff467a8e599609621887c6cdd
and replaces #19992 with documented nagfor option -unsharedrts, not
-unsharedf95, as pointed out by ThemosTsikas
|
| | | | | |
| | | | | |
| | | | | | |
* Fix typos
|
|\ \ \ \ \ \
| | | | | | |
| | | | | | | |
DOC: Deindent some sphinx declarations to avoid warnings.
|
| | | | | | | |
|
| | | | | | | |
|
|\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | | |
MAINT: Misc `np.array_api` annotation fixes
|
| | | | | | | | |
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
Changed as `Array` does not live in the main `np.array_api` namespace
|
| | | | | | | | |
|
| | | | | | | | |
|
| | | | | | | | |
|
|\ \ \ \ \ \ \ \
| |_|_|_|_|_|/ /
|/| | | | | | | |
MAINT: Removed the `cdoc` directory
|
|/ / / / / / / |
|