Stavros Macrakis <macrakis at alum.mit.edu> wrote:
On Thu, Dec 10, 2009 at 4:34 PM, R Fateman <fateman at cs.berkeley.edu> wrote:
BTW, the stack overflow issue is as far as I can tell completely orthogonal to the OS type; it is (or was, if it
has in fact been fixed) a Lisp implementation (ECL) problem.
Don't be too sure about this. My experience is that reliable
stack-overflow detection even within a single Lisp implementation is
very dependent on the was the particular low-level stuff in the OS
maintains guard pages. Not all processors/memory architectures do as
good job as one desires, and some extreme kinds of stack overflow are
very hard for the Lisp implementation to field reliably, at least
without slowing the routine linkage to make checking more reliable.