Why no tail calls from generators?
# Bergi (8 years ago)
Joe Gibbs Politz wrote:
What is the rationale for explicitly disallowing PTC in generators? (Or, in the case of proposals like STC, why not allow users to opt in to TCO in generator bodies?)
I'd guess that generator functions are already complicated enough, and
the next() call that at least has to wrap the result in {value: …, done:true} would always have to stay on the stack.
However, it would be possible to do tail-recursive calls in generator
functions via
return yield* f(…)
Kind , Bergi
I tried searching the archives and a bit around the Web, but couldn't find a place where this was written down; I may just need a pointer.
What is the rationale for explicitly disallowing PTC in generators? (Or, in the case of proposals like STC, why not allow users to opt in to TCO in generator bodies?)
Specifically, I'm referring to step 5 in IsInTailPosition that returns false ( www.ecma-international.org/ecma-262/6.0/index.html#sec-isintailposition) if the body is a generator body.
Thanks! Joe P