MAINT: 1.11.4 backports #19543
MAINT: 1.11.4 backports #19543
Conversation
tylerjereddy
requested review from
perimosocordiae and
rgommers
as code owners
There was a problem hiding this comment.
These backports LGTM, thanks Tyler. The CI failures (linter, and the doc build issue from gh-19500) are worth fixing.
The Poetry loongarch thing is worth fixing and not directly related to numpy 2.x changes, but yeah it's not super high on my TODO list and I think it can wait until 1.12.0
|
Thanks for mentioning #19513! I think I'm a bit confused by this statement:
If the CPython upper bound is unpinned, then why would that issue have to wait until 1.12 or 1.13? |
|
Yeah I think that links the wrong issue. Not sure why gh-19513 would not be closed either now or when |
Did you mean gh-19189 Tyler? |
|
Yes, I think I've got the links related to poetry stuff fixed up now. |
|
The CI is green here, but I'm working on the |
* set idx_dtype explicitly in dia_array.tocoo * avoid 64bit test_base for 32bit behavior
* fix pow method for sparrays with power 0 A**0 means every entry is 1, not just the originally non-zero entries. So, the A**x function changes the nonzero structure to dense in a discontinuous fashion near x=0. There is a question then as to whether this method should return a sparse array with all entries 1, or return a dense array, or raise an exception with a workaround described. This PR implements raising an exception for `n` a zero valued scalar. The message (and doc_string description of `n`) says if they want to compute A**x for x==0, they should special case that value to use `np.ones(A.shape, dtype=A.dtype)` * Update scipy/sparse/tests/test_array_api.py Co-authored-by: Jake Bowhay <60778417+j-bowhay@users.noreply.github.com> * change from ValueError to NotImplementedError and update test * Thanks for these suggestions:) Co-authored-by: Julien Jerphanion <git@jjerphan.xyz> * fix error message --------- Co-authored-by: Jake Bowhay <60778417+j-bowhay@users.noreply.github.com> Co-authored-by: Julien Jerphanion <git@jjerphan.xyz>
…9364) * MAINT: stats: update UNU.RAN submodule * TST: stats: add a test for DAU * TST: stats: add rtol,atol params to check_[cont|discr]_samples
… to bounds for method 'trf' Essentially reverts commit ba761d8 Closes scipy#19351
Closes scipygh-19357 Closes scipygh-19387 [skip circle]
…y#19408) * Fix typecasting problem in lil_matrix truedriv * Added test --------- Co-authored-by: Dan Schult <dschult@colgate.edu>
* update the SciPy 1.11.4 release notes following a series of backports [skip cirrus]
* attempt to fix up linter complaints related to `isinstance` usage for type checking [skip cirrus]
* DOC, MAINT: Bump CircleCI Python version to 3.11 * Add pin for myst-nb on CircleCI config to fix doc build issue [skip azp] [skip cirrus]
* update the SciPy 1.11.4 release notes following more backport activity [skip cirrus]
* temporary workaround to allow SciPy docs to build on `main` with Python `3.11` until an upstream fix for scipygh-19228 is available * I thought about guarding this filter by Python version, but didn't seem worth it, but can change that if really desired * locally, `python dev.py doc -j 32` has a `0` exit code/succeeds with `3.11` [skip cirrus] [skip actions]
* update SciPy 1.11.4 release notes following more backport activity [wheel build]
tylerjereddy
force-pushed
the
treddy_1_11_4_backports
branch
from
126a3c6
to
4f72d10
Compare
|
Updated the release notes to include the poetry fix/issues, and since CI was already green, now doing a trial run of wheel builds, including Cirrus, before more formal release process starts. |
Backports included so far (let me know if some are missing, I did a manual sweep for bug fixes though):
Currently not included are the
pyproject.tomlchanges related to gh-19189 and gh-19437--IIUC, these may be delayed until we unpin NumPy versions more broadly in the NumPy 2.x support release (which may be1.13.0for us, sometime a few months after1.12.0). Those shouldn't be confused with unpinning the CPython upper bound, that's already merged in before this PR.The full test suite and doc build passed locally on x86_64 Linux, but I'll skip Cirrus CI for the moment.