Re: ANN: Etiquette: a protocol construction tool

* "Peter" == Peter Seibel <[email protected]> writes:

Peter>  Another variant (which worked quite successfully in a message
Peter>  processing system I worked on in Java) is to use a fixed
Peter>  number of threads to provide an asynchronous interface to
Peter>  things that are fundamentally blocking. Thus you have one
Peter>  thread devoted to running your state machines (or possibly N
Peter>  threads, where N is the number of CPU's in your machine) and
Peter>  then you use one thread to deal with each place you have to
Peter>  blocking i/o. For instance, if you don't have select(), either
Peter>  because it's not exposed up through Lisp or because the OS
Peter>  implements it badly[1], or because you're dealing with some
Peter>  other API that doesn't expose itself as a filehandle, such as
Peter>  doing queries against a database, you can instead spin up a
Peter>  thread that accepts requests to do i/o on a queue and does the
Peter>  i/o as fast as it can.

This is actually an important point: in general we can't rely that the
underlying transport will use pollable machinery.

So it seems there is a number of ways how lightweight multiplexing of
protocols can be achieved. I'll need to study the issue more
thoroughly. But from the responses I see that this feature is
desirable, hence it is now placed on my to-do list.


--
  Eugene