Matthew Barber

So I wrote #497 to add what I think is obvious and uncontroversial: 1. `__complex__()` for complex arrays. 2. unrestrict casting with a different kind for all respective casting methods,...

On the 20th Oct meeting, no one opposed to going beyond unrestricted casting and explicitly support (most) cross-kind casting, emulating how Python's own builtins' cast to different kinds just like...

Well we currently git tag releases, as opposed to have branches (no idea how git works here, but that's an important distinction I'd gather). As stated above, I'm not sure...

> What do you mean "meta-standardising"? Ahah, like standardising how we write the Array API standard :sweat_smile: The effort wouldn't just be in fixing the signatures, but making sure the...

> but how are you actually planning to package the stubs into a package? I hadn't worked it out yet, but I'm sure it's possible with pulling repos-at-tags as package...

> You seem to be working off the assumption that the current status quo (signatures inside of the Sphinx directory for the current draft version only) is a good one,...

Will submit a PR fixing this

I didn't realise CI has 3.6 and 3.7 jobs. I see https://github.com/vaexio/vaex/pull/1802, so assume these versions want to be generally maintained for? I have used f-stirngs in this PR, although...

Forgot to update some refs of `size` (i.e. use `size()` instead), so made the necessary updates and wrote a test. Found a pre-existing bug with `Column.get_chunks()` that now has some...

> > pyarrow.lib.ArrowInvalid: Dictionary array invalid: Invalid: First or last binary offset out of bounds > > This seems like a legit issue :/ I wasn't even scrolling down the...