Currently I am working on making a BBS software into a MULTI-TASKING
software in the case of allowing multiple users on several nodes at once
-- aka Multi-Node! <G>  

        My current idea is to just have a repeat until statement around port
checking routines.  Is this the best way to do this?  My understanding
would be to make "processes" but how would a sample code go?  Please
send all your ideas to the newsgroup for expansion.  I have all the comm
routines done I just need to get this part added in before I go any
further.

- tasker.zip    to find at  Simtel /msdos/turbopas
- mtask20.zip   to find at  ftp.garbo.uwasa.fi/pc/turbopa45
- multi020.zip  to find at  ftp.garbo.uwasa.fi/pc/turbospec
or a newer project at
http://www.eikon.e-technik.tu-muenchen.de/~et_514/process.html

Maybe you find some more links at the (good) pascal-page
http://www.cs.vu.nl/~jprins/tp.html
( look under game-programming )

I didn't try them all, because I use since 4 years for Borland-Pascal
under Real-Mode-DOS the commercial RTKernel-Pascal
multitasking-realtime-library (also available for C), look at  

http://www.on-time.com

With simple DOS I can use our old XT's, AT's and 386's for automation,
measurement, machine-control etc.

Under DOS you have in your pascal-program preemtive multitasking (=threading),
real-time-capability, semaphores, task-priorities, no DOS-Reentrance problem,
interrupt-support, unlimited number of tasks, IPX and Serial units and .........

The product is not cheap, but if you (after some time) understand the
multitasking-concept you can spend your (precious) time only for
developing the (critical) applications (and they run fine).

Good luck

On Sun, 09 Mar 1997 00:49:21 -0500, Joe Radkowski <fjr...@psu.edu>
wrote:

We've written a database intensive, real-time-protocol PC application
for machine control and data capture, using a co-operative
multi-tasking approach (using the paradox engine - which is *not*
re-entrant), using a simple round-robin scheduler to achieve the
multi-tasking.  It worked fine, and was surprisingly quick.
*However*, the code was difficult to understand and even more
difficult to maintain/develop later.

Having taken a deep breath and purchased RTK, we've never looked back:
coding is much simpler, making it easier to understand/support the
product, and the additional features, such as IPX communications have
proved extremely valuable.

Bottom line: if it's a serious app which you are going to support
long-term (>1yr), do yourself a favour - effort now means less effort
later!