dan sinclair

This currently leads to some confusing spec text like: > then that operation may additionally return either the infinity with the same sign or the largest finite value with the...

This maybe a good place to start introducing `undefined value` as the undefined value for a f32 or f16 is `infinity or any valid value`. We then use `undefined value`...

Suggestions, some may not be easily done, not sure. * [x] Inline single rule elements * [x] Combine duplicate rules * [x] Right recursive rules can become `*` rules *...

* [x] Add a blank line between rules * [x] Rules with a `post.unary_expression` should use them * [x] Drop the terminals? # Rules with a `post.unary_expression` should use them...

Also `static_assert`. I wasn't sure about `struct` which was in that section as well so I kept it as a keyword for now.

We can always add extended forms later if they're requested. We can't remove them if they aren't useful. The benefits I see to `vec4` are: * Consistency, it's the same...

What was the request from the internal team? That it's too hard to type? Not the same as other languages? What is their issue that is trying to be solved...

I knew the power of 2, but 2^0 is 1, so that one held. I was looking in the Memory Layout section but that second point you quoted isn't there,...

I'd almost want it the other way, in Structure Member layout have a bulleted list of * Must be power of 2 * Must be n = k x RequiredAlignOf...

Actually, re-reading the `location` section, this isn't possible as you can't apply `location` to a struct. Only `numeric scalar` or `numeric vector`. So, there is no type we have that...