Subj : Yielding timeslices
To : David Noon
From : Lee Aroner
Date : Wed May 23 2001 09:59 am
DN>> Of course, well written native OS/2 code eliminates this
>> timeslicing problem completely.
LA> Just as well written native DOS code will as well. My DOS apps
LA> barely register on any of the various utilization checkers, and
LA> this despite using a polling type scheme for keyboard input.
DN> If your program is polling, it is not as well designed as a native OS/2
> program. The yielding of timeslices repeatedly is not as
> satisfying a solution as blocking completely within a
> device driver.
Of course it is not as well designed as an OS/2 application would
be, it's built to run under a lowest common denominator
cheeseball interrupt handler swiped from a guy who copied it from
someone else!
And in this case, the polling serves other purposes...the routine
that surrenders timeslices also has an installable call of it's
own which allows the dynamic insertion of a chain of secondary
tasks that provide internal multitasking. All my apps can print
in background, for example. Not as neat and nifty as OS/2
certainly, but not bad for DOS.
And, for DOS apps, my proggys are *very* well behaved, and that,
in the end, is what counts.
LRA
-- SPEED 2.01 #2720: "He created OLD fossils!" "Riiiigghht!"
--- Maximus/2 3.01
* Origin: Top Hat 2 BBS (1:343/41)
