[MLton] stack-allocated references

[Jake Donham]

skaller <skaller at users.sourceforge.net> writes:

> Of course there is a bound per program, which a whole program
> analyser like Mlton can no doubt be coaxed into calculating, and of
> course there is no universal bound for all programs.

Sure, of course--I meant an implementation limit.

> What happens to Cheng's algorithm if instead of only the top
> stacklet being mutable, the top N stacklets are mutable, for some
> fixed N?

Well, from the point of view of the realtime bound, you might as well
just make one big stacklet of the N mutable ones. You have to process
the mutable stacklet(s) when you shut down the collector, so their
size (and the size of other units of GC work of course) determines the
realtime granularity.

Jake