Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

WIP,API: Enable full Numpy2 compat #15897

Closed
wants to merge 11 commits into from
Closed

WIP,API: Enable full Numpy2 compat #15897

wants to merge 11 commits into from

Commits on May 31, 2024

  1. API: Change Scalar to raise for invalid Python values 

    This aligns with NumPy, which deprecated this since a while and
raises an error now, for example for `Scalar(-1, dtype=np.uint8)`.

Necessary to ensure we give the right errors when promotion doesn't
up-cast based on values.
    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    704771d View commit details
    Browse the repository at this point in the history

  2. TST: Skip test when running under buggy NumPy versions

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    fd27aa0 View commit details
    Browse the repository at this point in the history

  3. TST: NumPy now prints np.int8(1), keep using old printing in tests

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    ecf1558 View commit details
    Browse the repository at this point in the history

  4. TST: Silence cupy warnings (TODO: not sure this is needed)

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    a5a52ba View commit details
    Browse the repository at this point in the history

  5. TST: NumPy is deprecating "a" as a dtype, just skip it

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    89162d1 View commit details
    Browse the repository at this point in the history

  6. MAINT: Use int8(-1) as default categorical as np.uint8(-1) fails

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    21d25aa View commit details
    Browse the repository at this point in the history

  7. MAINT: Adapt to e.g. uint8(-1) failing now. Mostly in where. 

    Note that pandas `where` seems to promote the Series based on the
value even with NumPy 2.  This was never copied by cudf (i.e. an
outstanding issue)
    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    b17af57 View commit details
    Browse the repository at this point in the history

  8. MAINT: avoid can_cast(pyscalar, dtype) NumPy 2 refuses it (for now)

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    9e5d611 View commit details
    Browse the repository at this point in the history

  9. MAINT: Adapt numerical promotion to NumPy 2 and Pandas 2.2 

    Pandas keeps using weak promotion even for strongly typed
"scalars" (i.e. 0-d objects).
This tries to (mostly) match that, but there may be better ways to
do it.  I am having difficulty to think of the best way though.
    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    314d047 View commit details
    Browse the repository at this point in the history

  10. MAINT: Adapt to NumPy weak-promotion in comparisons

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    d1c7224 View commit details
    Browse the repository at this point in the history

  11. TST: Add test for equality, this test fails as of now!

    @seberg
    seberg committed May 31, 2024
    Configuration menu
    Copy the full SHA
    d69d9cb View commit details
    Browse the repository at this point in the history