Almann Goo

I don't think `TIME` should sort together with anything--it has no point in time meaning. I am even suspicious that even `DATE` would make much sense here since that isn't...

> To be clear, only this commit was intended to bed in this PR. I will fix tomorrow. > > [7632092](https://github.com/partiql/partiql-spec/commit/76320926ebea683f9f0fb99bb554ecc794759a5e) Thought so--once you rebase, I will happily :ship:.

Good point, I definitely think we should be clearer about what levels of SQL compliance in the spec itself.

> 1. Nested according to the lexical order of LET clause? I believe this is the form we probably desire. This is basically what we get with *correlated joins* in...

> 2. All let-bindings are available to all let-expressions? I *don't* think this is a good candidate for this proposal for a couple of reasons: * It is inconsistent with...

Specifically this caused a regression with #6.

It is a very reasonable suggestion and I think ANTLR4 is probably the strongest contender here, though I do want to keep our options open (e.g. maybe some form of...

Yeah, this does seem a bit more involved than it would seem worth one's time--did anyone actually do fuzzing on the ARM implementation? Mostly curious to see how platform limited...

FWIW, I've run into cases where this would be useful. In amzn/ion-rust#217 I manually branch around the ignore case, but the test itself is not ignored.

A lot of Python packages eschew that many characters, so my inclination is to flatten and make it around six-ish characters. Another option is `ionpy` (in the spirit of your...