[MLton-user] Destructive update

[Stephen Weeks]

> isn't it true that the first thing is to say how much RAM a process
> can use, and then, given that number, what kind of GC to do (via
> stop-and-copy vs. 2-space)?

Yes.

> I don't see why it is always using the physical to determine the
> latter.

Perhaps I misspoke -- the physical amount of memory is *not* used to
determine the type of GC.  It is the amount of memory the process is
told it has, as determined by ram-slop.  The s->ram used to compute
the ratio that governs the type of GC

        ratio = (float)s->ram / (float)live;

is the same one that was set at program start via

        s->ram = align (s->ramSlop * s->totalRam, s->pageSize);