Sounds good. I think that we should also try to lift the restriction that <op> operands are <template> and make it so that expr ::= (<op> <expr> <expr>) works.
I think that the current semantics don't have a first-class notion of booleans. enabled_if just wants a string that looks like one. So if that's true, (> true false) should be accepted.
Sounds good. I think that we should also try to lift the restriction that
<op>
operands are<template>
and make it so thatexpr ::= (<op> <expr> <expr>)
works.@emillon do we allow things like
(> true false)
?