Jake Vanderplas

I suspect this is working as intended, since `spsolve` is a lower-level function, though we should do a better job of documenting its requirements.

I don't think we'll ever automatically take care of this, because deduplication is a relatively expensive operation (even detecting whether deduplication is necessary is expensive!), and people calling a low-level...

Thanks for the report! Lowering contiguous slices to `lax.slice` was an optimization we made a while ago, and at the time we scoped the problem to contiguous slices for simplicity....

Agreed, that's why I marked this as a performance-related enhancement.

Can you say more about why you want to enable this test? It's only testing `dot_general` with `preferred_element_type` – we do have other coverage for `dot_general` without this argument.

Could you sync against the updated main branch and resolve conflicts? Thanks!

It looks like this breaks some tests in `lax_numpy_reducers_test.py`. You should be able to repro by running ``` JAX_NUM_GENERATED_CASES=90 pytest -n auto tests/lax_numpy_reducers_test.py -k testCumulativeSum ``` It looks like this...

Before we merge this, could you squash your changes into a single commit? See https://jax.readthedocs.io/en/latest/contributing.html#single-change-commits-and-pull-requests Thanks!

Typically the example comes below the description of the function parameters. I think in this case it would be best to rewrite the docstring from scratch. I'm doing some similar...

Oh, and please don't open a new PR if possible – it's useful to keep all the conversation in one place. The failing CLA is probably due to whatever email...