[MLton] Re: [Haskell-cafe] fastest Fibonacci numbers in the West
Thu, 27 Jan 2005 11:28:14 -0800 (PST)
On Thu, 27 Jan 2005, Henry Cejtin wrote:
> setrlimit can only increase the soft limit and only up to the hard limit.
> MLton uses its own stack for the ML stack, but there isn't much you can do
> for the C stack that C code runs in. (We could make the code use stack space
> allocated some where else, but there is no way to make a GC happen and the
> stack grow if you run out.)
Why not? Couldn't you have some guard pages at the end so that when they
were hit the signal handler would trigger a re-allocation and then a GC?