On 08/16/2013 12:57 AM, Markus Mottl wrote:
> Hi,
> 
> I just wondered how much we can rely on current OCaml-semantics where
> context-switches are impossible as long as there are no allocations.

I wrote a little test program and this is not true for bytecode -vmthread, but seems to work for bytecode -thread and native -thread:
$ ocamlfind ocamlc race.ml -package threads -vmthread -linkpkg -o race && ./race
started
Check OK
worker 2 OK
worker 3 OK
worker 4 OK
worker 5 OK
worker 7 OK
worker 8 OK
worker 9 OK
worker 10 OK
worker 11 OK
worker 12 OK
worker 13 OK
worker 14 OK
worker 0 OK
worker 1 OK
worker 6 OK
worker 15 OK
Array sum inconsistent: -33


> 
> Anyway, is it considered reasonably future-safe to write code of that
> sort?  I'd rather not have new compiler optimizations, etc., interfere
> with these assumptions in the near future.

I'd consider this very fragile as you'd need to know the implementation details of every function you call (or only calling your own safe functions). I was surprised to find out that even Array.fold_left creates a temporary ref for example.

Best regards,
--Edwin