Tim Millar

The docs suggest that `NUMBA_DISABLE_JIT` should be ignored by `@guvectorize` as it only only affects `@jit` and `@njit`: https://numba.pydata.org/numba-doc/dev/user/troubleshoot.html#disabling-jit-compilation

In the interest of getting things moving, we could: - Stop running the test suit with `NUMBA_DISABLE_JIT=1` - Only use `NUMBA_DISABLE_JIT=1` on tests that don't hit `guvectorize` (e.g., `test_pedigree.py`) We're...

I think the way forward is to annotate all of the guvectorize functions with `# pragma: no cover` again.

It can be controlled with an xarray setting: https://docs.xarray.dev/en/stable/generated/xarray.set_options.html#xarray-set-options

@jeromekelleher your interpretation looks correct to me. We should also test for a case where `-2` should appear in an INFO field (e.g., with length `A` or `R`): ``` ##INFO=...

Hmm, not sure what to do about coverage here

This still runs tests with the jit disabled for two files: - `sgkit/tests/test_pedigree.py` - `sgkit/tests/io/vcf/test_vcf_writer_utils.py` Both of these files have lots njit functions without guvectorize. This is a bit fragile...

I've just come across this issue when trying to calculate the number of unique alleles at each site. I don't actually think this should be an issue if we are...

I noticed that the dtype for `variant_allele` can be bytes or objects (`'O'`). This makes it unclear what a generic 'non-allele' value could be. Is there a particular use-case for...

That makes sense, I'd guess that checking for null values in an object array should include checking for any of `b'', '', None` to be safe. But I think we...