SLED and mlton
Tue, 18 Sep 2001 10:20:19 -0500
We hope for reasonable defaults to and I am trying to figure out why on some
machines, for some people, some of the time they don't work. What is
supposed to happen is that on start up MLton looks to see how much physical
RAM the machine has and sets the maximum it will ever ask for to some
fraction of that (currently 85%). Clearly this isn't really right if you
have other big processes running at the same time, but should be a reasonable
first wack. One problem is that if we ever ask for memory under this amount
and the OS says no, we exit instead of properly backing off.
Is the failure you observer repeatable? If MLton is just exiting with a
message about an mmap failing then it is just the known problem, and this can
be `fixed' by having more swap. If it is something else (like it starts
swapping forever) then I would like to know how big MLton is trying to get
and how much RAM you have.