API: Add new `np.exceptions` namespace for errors and warnings by seberg · Pull Request #22644 · numpy/numpy

seberg · 2022-11-22T12:41:01Z

This adds the namespace and populates it with the main public exceptions and warnings that we have. I skipped RankWarning for now, because we have two (and only one is in the main namespace).
The UFuncTypeError is currently effectively private, so did not move it quite yet (this can be a followup).

Based of gh-22619 and thus creating as draft. I put 1.24 as version here, but most likely this will need updating.

The main point of this is really to simplify adding of new exceptions in the future, but I think the module doesn't really hurt even for 3 errors.

Mailing list thread where we discussed adding this module: https://mail.python.org/archives/list/numpy-discussion@python.org/thread/IZJQTPQBKNQGK4OGWSTREQWKGMQDSFKM/

seberg · 2022-11-22T12:52:07Z

numpy/exceptions.py

+    .. warning::
+
+        This warning should not be used, since nose testing is not relvant
+        anymore.


Added this warning, and skipped documenting it... It seems like a remnant of a time past.

seberg · 2022-11-24T12:03:11Z

Can we please move this (and the other PR there) forward? I (+do not see) potential for controversy or code complexity here, but:

I need this to introduce new errors
I need new errors to have a chance of fixing up the gigantic mess that has been around unfixed for years, that our == FutureWarning is. (The FutureWarning is good, but its useless if we don't know how to finalize it)
I need THAT one to be able to paint a full picture of the option to make API: No timedelta/datetimes are not integers and do not promote #22568 happen.
And that again, would at least help a little with clearing up the whole mess that find_common_type is.

I bet there is more, but my monkey brain can only hold 3 things at once.

mattip · 2022-11-24T15:23:16Z

numpy/exceptions.py

+    AxisError       Given when an axis was invalid.
+    TooHardError    Error specific to `numpy.shares_memory`.
+
+"""


Do we have autosummary elsewhere in a module docstring? This feels out of place to me.

Hmmm, fft and polynomials used it. But I am not sure the autosummary here is actually useful.

I still think this belongs somewhere else.

Where? Maybe @rossbar can just say.

mattip · 2022-11-24T15:25:33Z

numpy/lib/npyio.py

+            starting with '#' or as specified via `comments`) are not counted
            towards `max_rows`.
    quotechar : unicode character or None, optional
        The character used to denote the start and end of a quoted item.


While I agree with the whitespace cleanup in files you are changing anyway, I don't think this one should be part of this PR.

Oh, not even an actual change left in the end here... (set_module is imported via overrides here, which seemed OKish.)

charris · 2022-11-24T15:26:27Z

Note that this is still a WIP, might want to change that so it can be merged.

mattip · 2022-11-24T15:28:52Z

This is still draft status. For other reviewers: it has 3 components:

moving two warnings, another unused warning, and two exceptions into an exception namespace
adding a new _util namespace and moving set_module to there
cleaning up whitespace

seberg · 2022-11-24T15:38:18Z

I have a separate PR for moving the _set_module in gh-22619, the reason for the draft status.

Yeah can pull out the whitespace changes there, they were not really intended, just auto-stripping by the editor.

mattip · 2022-11-29T21:34:42Z

Is this ready to go? I quite like it.

This means moving ComplexWarning, TooHardError, and AxisError.

Both should now live in the "exceptions" module.

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

seberg · 2022-11-30T11:09:34Z

Should be ready to go now, was missing a rebase on main still.

mattip · 2022-11-30T12:21:33Z

numpy/exceptions.py

+
+

Suggested change

mattip · 2022-11-30T12:23:54Z

numpy/core/numeric.py

    convolve : Discrete, linear convolution of two one-dimensional sequences.
    multiarray.correlate : Old, no conjugate, version of correlate.
-    scipy.signal.correlate : uses FFT which has superior performance on large arrays. 
+    scipy.signal.correlate : uses FFT which has superior performance on large arrays.


This is tripping up the linter since the line is 85, not 80. But I think it is OK.

Co-authored-by: Matti Picus <matti.picus@gmail.com>

mattip · 2022-11-30T21:08:08Z

Thanks @seberg. There is still the documentation issue, but lets leave that for a future PR.

Xref numpy#22644 and numpy#22707 Update modules.pyi

github-actions bot added the 30 - API label Nov 22, 2022

seberg commented Nov 22, 2022

View reviewed changes

mattip reviewed Nov 24, 2022

View reviewed changes

seberg force-pushed the exceptions branch 2 times, most recently from b7dfc6e to a82a155 Compare November 24, 2022 16:04

seberg marked this pull request as ready for review November 30, 2022 07:52

seberg force-pushed the exceptions branch from a82a155 to 20c5af2 Compare November 30, 2022 07:53

seberg added 7 commits November 30, 2022 12:07

API: Add new exceptions module and move exception exposed via numeric

b0f318b

This means moving ComplexWarning, TooHardError, and AxisError.

MAINT: Move ModuleDeprecationWarning and ModuleDeprecationWarning

1d31511

Both should now live in the "exceptions" module.

TYP: Modify typing stubs for new np.exceptions module

86029d0

DOC: Document exceptions and warnings in the refguide

7dce375

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

DOC: Add release notes for np.exceptions namespace

f2c5b1d

DOC: Update messages to NumPy 1.25 although I hope its 2.0 :)

7b1ee55

MAINT: Fixup new ufunc AxisError use to use numpy.exceptions

8cf1f69

seberg force-pushed the exceptions branch from 87d2ad9 to 8cf1f69 Compare November 30, 2022 11:08

MAINT: (meson) Add exceptions and expctions.pyi to pure python sources

e9e1d03

mattip reviewed Nov 30, 2022

View reviewed changes

numpy/exceptions.py Outdated

Comment on lines +65 to +66

Copy link
Copy Markdown

Member

mattip Nov 30, 2022

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Suggested change

mattip reviewed Nov 30, 2022

View reviewed changes

Update numpy/exceptions.py

4539ecd

Co-authored-by: Matti Picus <matti.picus@gmail.com>

mattip merged commit 9896b46 into numpy:main Nov 30, 2022

seberg deleted the exceptions branch November 30, 2022 21:58

BvB93 added a commit to BvB93/numpy that referenced this pull request May 8, 2023

TYP: Add the np.exceptions namespace and add DTypePromotionError

c6fdee0

Xref numpy#22644 and numpy#22707 Update modules.pyi

DavidPost-1 mentioned this pull request Nov 27, 2024

AttributeError: module 'numpy' has no attribute 'ComplexWarning' nkarast/PyNAFF#9

Closed

Uh oh!

Conversation

seberg commented Nov 22, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

seberg commented Nov 24, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

charris commented Nov 24, 2022

Uh oh!

mattip commented Nov 24, 2022

Uh oh!

seberg commented Nov 24, 2022

Uh oh!

mattip commented Nov 29, 2022

Uh oh!

seberg commented Nov 30, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mattip commented Nov 30, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

seberg commented Nov 22, 2022 •

edited

Loading

seberg commented Nov 24, 2022 •

edited

Loading