Damien Doligez

`tools/ci/actions/runner.sh` needs to be updated too (for the `basic-compiler` action).

@gasche, sanitizers won't help you in this case, the leak seems to be in the heap, probably because the major GC doesn't auto-trigger often enough. I'm looking into it.

I tend to agree with @jeremiedimino : `ocamldep` should not output any dependencies if there's an error. But there's not telling what strange things build systems will do, so a...

@gasche do you think this could be a "good first issue"?

In general, it would be good to have every flag come in pairs to disable/enable. That makes it easier to do all kinds of build-system hacks, and it also makes...

Still a good "first issue".

> I agree that ocamlopt could generate code that calls `caml_alloc_shr` instead of `caml_alloc` for large allocations. In the old days, the block was filled using `caml_modify`, so unit-initialization of...

`gnu11` is C11 with gnu extensions, right? That looks reasonable to me.

- [x] Revisit the capping of work-to-do. https://github.com/ocaml-multicore/ocaml-multicore/blob/a3832eeaae08ec103d2b4377b5624c081e2782c2/runtime/major_gc.c#L411 https://github.com/ocaml-multicore/ocaml-multicore/blob/a3832eeaae08ec103d2b4377b5624c081e2782c2/runtime/major_gc.c#L420 In the sequential GC, the work was capped because you can't do more than 1/3 of a GC cycle in one...

About `shared_heap.c` - [x] This is relying on the initial value of mapped memory: https://github.com/ocaml-multicore/ocaml-multicore/blob/a3832eeaae08ec103d2b4377b5624c081e2782c2/runtime/shared_heap.c#L172 I'd rather initialize the first pool explicitly like the others. - [x] `Assert` vs `CAMLassert`:...