[MUD-Dev] Re: Multi-threaded mudding (was a flamefest)

[clawrenc at cup.hp.com]

In <199705020501.AAA14692 at dfw-ix9.ix.netcom.com>, on 05/01/97 
   at 10:04 PM, "Jon A. Lambert" <jlsysinc at ix.netcom.com> said:

>> From: Jeff Kesselman <jeffk at tenetwork.com>

>The implication is, if my definition of event-driven systems is
>correct, program control MUST be external to the procedural code of
>the events  for a server to be truly "event-driven".

>Can it be done with a single-threaded 'dispatcher/monitor'? 
>Certainly. If OS multi-threading is superior to software time-slicing
>on your platform then it would follow, that a multi-threaded
>event-driven server would outperform a single-threaded event-driven
>server regardless of  how many CPUs your processor has.  

>Are there any faults with this logic?

None here for the general case.  The main reason to implement
appilication threads (I think MS calls them "fibers") is that you can
get a performance benefit and design simplification for co-operative
threading (eg explicit yields).

--
J C Lawrence                           Internet: claw at null.net
(Contractor)                           Internet: coder at ibm.net
---------------(*)               Internet: clawrenc at cup.hp.com
...Honorary Member Clan McFUD -- Teamer's Avenging Monolith...