KC Sivaramakrishnan

> Are you saying that it is to be expected that code that relies on Obj can now crash the GC when it did not before? No. That's not the...

This is tricky. See #719. Closing this since #719 is more descriptive.

There is also the bigger question of what, if anything, should be done regarding other mutexes in the runtime system introduced by Multicore OCaml. Given that Multicore OCaml will (should?)...

Thanks for the report and the reduced expect test. Please can you describe the steps to reproduce the segfault with the new test?

@Engil Since we've merged #629, this issue should be fixed. Am I right?

> I'm trying to reduce that. Thanks @emillon!

IIRC stock OCaml never returns memory back to the OS. This is not ideal. Perhaps the choice of how much memory to hold should be tied to the space overhead...

In Multicore OCaml, large allocations (> 128 words) go through `malloc`, and are `free`'d when swept.

> have a run-time hook in Domain.spawn that "remaps" and only triggers the first time the runtime goes beyond 1 domain. We have a `Domain.at_first_spawn` function that runs at the...

@Gbury Thanks for clarifying how AFL fits in the compilation.