Open
Description
The title says it all: I have written a grader which works in batch mode (when executed via learn-ocaml grade
) but fails in the browser (Firefox) with a Stack_overflow
exception. This happens when grading the solution of my exercise nondet_monad_cont
, which currently is sitting in a pull request in learn-ocaml-corpus
. The second half of Question 4 (laziness) is where the grader fails.
Is it possible that the compilation to JavaScript does not do tail call optimization?
Is it possible to somehow ensure that the stack limit is the same under both environments (batch mode and in-browser mode)?
Is it possible for each exercise to indicate what stack size it needs?