Is the idea that this is for situations you can't set `ulimit -s` ?
This seems like a lot of work when you could just increase the stack size instead.
How well does that work on code running in an interpreter? Which interpreters have a 1:1 mapping between language-being-interpreted stack frames and native stack frames?
How well does that work on code running in an interpreter? Which interpreters have a 1:1 mapping between language-being-interpreted stack frames and native stack frames?