Null is a primitive type - does it need a lib?

[crusso]

TIL Null implements == and !=, but why bother? (o == null does not work for options o : ? T, only for o : null.

[crusso]

@rossberg?

[nomeata]

Just to keep the flow of unhelpful comments going: how about we ditch the Null type and have null : ?empty? (I may have asked this before and have forgotten the reason)

[crusso]

I've forgotten the answer, but why is the better and not just uglier?

[rossberg]

Wait, why does option == null not work? That would be a bug, given that Null is a subtype.

[nomeata]

I've forgotten the answer, but why is the better and not just uglier?

It’s maybe better because we don’t need to explain why there is a separte type called Null, why Null and ?t are related etc.

But I guess that ship has sailed anyways since we have null in Candid…

[crusso]

Wait, why does option == null not work? That would be a bug, given that Null is a subtype.

Why should that work? We don't derive equality functions for option (they are generic). Unless you want to special case that options of equality type admit equality, but then why not do that consistently for all equality types.

[crusso]

FTR, ordering on null is not supported (though equality is).

[rossberg]

Why should that work? We don't derive equality functions for option (they are generic).

Ah, right.