Portable Task Management

[Peter Rowell]

We have written what we *believe* is a fairly portable set of task
managment routines.  We are using setjmp and longjmp to save and
restore the register state.  There is a very small piece of environment
specific code (diddling the stack pointer(s) in the jmp_buf), but other
than that it seems to work quite well.

I know that there is probably at least one significant machine out
there that will cause us to either (a) write our own code to save and
restore the register state or (b) wish we had not gone down this road.

My question is: specifically which machines really ARE going to cause
us trouble?  I suspect it will be any machine where you can't have
arbitrary stacks lying around in memory, or where setjmp/longjmp (or
their equivalent) cannot be implemented for some reason.  I do not
personally know of such a machine.  I thought that Pyramid would cause
problems, but Hey! their setjmp man pages *tell* you how to do things
like this.

So, is there a Big Bad Wolf?  (and what is his net address?)

----------------------------------------------------------------------
Peter Rowell				peter at thirdi.UUCP
Third Eye Software, Inc.		(415) 321-0967
Menlo Park, CA  94025   		"Let's jmp in the deep end!" Me