[BUG] CI failure due to KASBA distances

[MatthewMiddlehurst]

Describe the bug

This is causing failures in multiple PRs. It is possible it may be tslearn related as we recently changed the bound for that due to an update?

Steps/Code to reproduce the bug

https://github.com/aeon-toolkit/aeon/actions/runs/19423271227/job/55582592734?pr=3116

https://github.com/aeon-toolkit/aeon/actions/runs/19419534167/job/55582544133?pr=3115

Expected results

Test passes

Actual results

FAILED aeon/clustering/averaging/tests/test_kasba.py::test_kasba_distance_params[distance7] - TypeError: bad argument type for built-in operation

Versions

No response

[Lucky-Lodhi2004]

The test runs on my local machine well.

What must be wrong

Certain versions of numba don't support named parameters and i see named parameters in /aeon/clustering/tests/averaging_ba_utils.py namely in shape_dtw_alignment_path, adtw_alignment_path, and soft_dtw_alignment_path

elif distance == "shape_dtw":
        return shape_dtw_alignment_path(
            ts,
            center,
            window=window,
            descriptor=descriptor,
            reach=reach,
            transformed_x=transformed_x,
            transformed_y=transformed_y,
            transformation_precomputed=transformation_precomputed,
        )
    elif distance == "adtw":
        return adtw_alignment_path(ts, center, window=window, warp_penalty=warp_penalty)
    elif distance == "soft_dtw":
        return soft_dtw_alignment_path(ts, center, gamma=gamma, window=window)

Converting these to positional argument might work, or we may check numba version in CI, as tslearn was updated recently.

[bhakunisagar512-ui]

Hi @MatthewMiddlehurst

I wanted to check whether this issue is already resolved or if there is an active PR addressing it.
If it’s still open, I’d be happy to work on a fix.

From the discussion, it seems the CI failure may be due to certain Numba versions not supporting named arguments in @njit functions, particularly in the KASBA alignment path utilities. Converting these calls to positional arguments (or otherwise handling version compatibility) might resolve the CI failures.

Please let me know:

• if this issue is already fixed or in progress, or
• if it’s okay for me to start working on it.

Thanks!