I don't see any need for a "part 3", but I thought I'd throw out an
interesting observation that just came up...

While we microcontroller people are starting to put more complex multitasking
schedulers into very small processors, the "big computer" folk are looking
at implementing SIMPLER forms of multitasking (ie ways of devolving tasks
into state machines.)  This is largely needed to scale performance for modern
computing tasks.  Seems that if you have something like a great big web
server, then using a full fledged process, or even a lightweight "thread"
for each of the expected tens of thousands of simultaneous tasks (each
lasting for only a couple of seconds) is just WAY too much overhead...

Interesting.

BillW