#: 11565 S10/OS9/6809 (CoCo)
    31-Jul-91  23:57:41
Sb: #11564-#Timer Thoughts
Fm: PaulSeniura 76476,464
To: PaulSeniura 76476,464 (X)

..

So we scratched our heads (again on Delphi -- no one was responding to CIS here
as I've been saying during this time).  I "discovered" F$SSWI on page 8-37 in
the Tandy L2 Tech Ref.  Tried to get people to find out if OS9 would swap a
task in (if sleeping or whatever) immediately if not sooner after a SWIn. Eddie
Kuns (who was very responsive as Jason was) said he thought it would since it
seemed such an important action.

And the watered-down boiled-down ideas are in that second Timer update article
I posted.  That's where we stand today, still in the research phase even before
we can write code to test it.  I *always* do this kind of research before
sharing even an alpha-test program with people.  I can't continue without some
solid facts whether our ideas have any merit whatsoever, or should we consider
other alternatives, or go completely back to the drawing board?

So now let me reexplain the SVC/F$SSWI method we'd like to try but have no idea
how to pursue testing it (I don't have a system-level debugger) or if it'd even
work with our assumptions.

First, the user pgm would do a SVC to inquire about the timer's resolution etc.
The user pgm can use this in the mathmatics of converting its timing to that of
the hardware/system.

The user pgm then sets up a computed-length delay, and tells OS9 to branch to a
routine (in its own area *or* to another subroutine module) when the timer
pops.  The user pgm may decide to sleep or to generate advanced data while
waiting for the delay to end; this shouldn't matter to the timer SVC itself.

..

There is 1 Reply.