>
> Simulate the whole day every minute like I was talking about.  It sounds
> like a horribly inefficient thing to do, but what else is the PIC going
> to be doing?  Also it should lead to small code, which is vital for you.
>
 I think you mean something like this:
  for x = 1 to 20     ; max. N^ of setpoint (in random order)
        If set_point_time(x) = present_time then
        buffer = third_byte_eeprom(x)    ; it said what  to do
        end if
 next x
 PIC_out_pin = buffer
and repeat this at least every minute .
I can't understand how to get a correct  refresh of the buffer every minute
 with the whole day min. by min. comparison.

> Although your way based on compare if greater isn't too bad either.  As
> someone else suggested, rather than do a full sort, just keep the table
> in order from the beginning.  Find the proper place to store enter a new
> set-point and move the other entries in the table up to accomodate it.
> Keeping the table in order should make it easier to use as well.

>
Yes. It's a good idea . I'll try it.

> Yes, that's another way to do it.  You'll need a way to subtract times,

I written my idea in that way :
buffer = 0    ;stay off until there is a match
prev_dist = 1440   ; 1  +   hh*60 + mm  (23*60+59 )
for x = 1 to 20       ;max. N^ of setpoint  (in random order)
      If present_time  > =  set_poin_time(x)     then
        dist = present_time - set_point_time
             if prev_dist > dist  then
                 prev_dist = dist
                         buffer = third_byte_eeprom(x)
                     end if
      end if
 next x
PIC_out_pin = buffer
and repeat this at least every minute.
In qbasic seems it work ; now i 'll try to translate in asm.