Closed
Description
Posting as a new issue rather than trying to coordinate with comments on various other PRs. In the community meeting earlier this week we discussed that we are almost ready, and @charris planned to branch maintenance/2.0.x today. We still have a few more things to be done though it looks like.
Here is what I'd propose we do today:
- Merge API: Make descr->f only accessible through
PyDataType_GetArrFuncs#25897 - Finalize and merge MAINT: remove
numpy.array_apimodule #25911 (@rgommers) - Revert API: Introduce
copyargument fornp.asarray[Array API] #25168 (because of__array__copy keyword changes incomplete #25916) (Done in BUG: fix handling of copy keyword argument when calling __array__ #25922) - Then build new nightlies (@rgommers) (Triggered manually by @mattip)
- Once the nightlies are up: ensure things work with using
npy_2_compat.hin SciPy (MAINT: shim for descr->f access scipy/scipy#20108), getting that merged and uploading new SciPy nightlies
After that (tomorrow?):
- Fix up API: Introduce
copyargument fornp.asarray[Array API] #25168 and get it merged again (@mtsokol and @ngoldbaum) - If blocking for 2.0.0rc1: get MAINT,API: Make metadata, c_metadata, fields, and names only exist on old-style dtypes #25802 in (@seberg)
- Last ABI-breaking PR,
still to be opened by @seberg: API: Restructure the dtype struct to be new dtype friendly #25943
Other things:
- Another pass over the release notes (@rgommers) (DOC: update and copy-edit 2.0.0 release notes #25921, DOC: 2.0 release highlights and compat notes changes #25937)
- May want to get API: Remove broadcasting ambiguity from np.linalg.solve #25914 in if there are no objections to that one
Everything else under the 2.0.0 milestone seems like it can wait.