Am Freitag, den 24.02.2017, 10:38 +0000 schrieb Richard W.M. Jones:
> (2) We have lots of existing code which does:
> 
>   CAMLlocal1 (v);
>   // some code here which might allocate and call the GC
>   v = ...
> 
> It's my understanding that such code might crash with the older
> versions of OCaml, because when it calls the GC it will find a local
> root which is initialized to 0.  We want to continue to support the
> older versions (back to 3.12 in fact), so should we change it all to:
> 
>   CAMLlocal1 (v); v = Val_unit;
>   // some code here which might allocate and call the GC
>   v = ...
> 
> ?  Interestingly none of this code has actually crashed in production
> as far as I'm aware.

Because it is correct for older OCaml versions - which do not follow
pointers that are not allocated by OCaml. Note that this is still the
default, but you can now change it with the -no-naked-pointers
configure switch, in which case such pointers lead to a crash when they
are run into by the GC (well, I'm not sure about a null pointers,
though). The -no-naked-pointers switch accelerates in particular
assignments (caml_darken called via caml_modify).

Gerd
-- 
------------------------------------------------------------
Gerd Stolpmann, Darmstadt, Germany    gerd@gerd-stolpmann.de
My OCaml site:          http://www.camlcity.org
Contact details:        http://www.camlcity.org/contact.html
Company homepage:       http://www.gerd-stolpmann.de
------------------------------------------------------------