[PEAK] Anybody using reactors?

[Phillip J. Eby]

At 06:58 PM 1/14/04 +0200, alexander smishlajev wrote:
>Phillip J. Eby wrote, at 14.01.2004 18:35:
>
>>>* ability to remove sheduled procedures from the laters list
>>I'm curious how this use case would be handled with 'peak.events'.  I 
>>haven't yet felt a strong need for it myself, obviously, else I'd have 
>>added it to UntwistedReactor.  Do you have a sample usage or kinds of 
>>usage that I could see, so I could see how I'd write it with peak.events, 
>>and make sure it's still tractable?
>
>we use it to handle timeouts: timeout procedure is registered with 
>callLater(), and we continue with normal data flow.  when expected event 
>arrives, timeout procedure (if any) is removed from the queue, with 
>possible resheduling at a later time.

Ah.  Okay, so something like:

while True:

     yield events.AnyOf(data.readable, scheduler.timeout(30)); 
src,evt=events.resume()

     if src is data.readable:
         # got something, act on it
     else:
         # timed out

would work, then?  Yielding on an 'AnyOf()' automatically discards 
callbacks received after the first one it receives from any of its 
inputs.  So, the above is a loop that waits for the 'data.readable' 
condition with a 30-second timeout each time.  It could also be written:

untilReadableOrTimeout = events.AnyOf(data.readable, scheduler.sleep(30))

while True:

     yield untilReadableOrTimeout; src,evt=events.resume()

     if src is data.readable:
         # got something, act on it
     else:
         # timed out

('events.IScheduler.sleep()' returns a relative delay object that is reset 
each time it's used, while timeout() returns a fixed deadline.  So, a 
sleep() object is reusable, while a timeout() would have to be created each 
time through the loop unless a fixed deadline were desired.)