LoopingCall (and/or everything else) dies when time goes backwards

New submission from colin <[email protected]>:

When time goes in reverse, LoopingCall ceases to execute until time catches up again via ntpd or whatever. 

This might effect the entire twisted scheduler.

{{{
#!python
from twisted.internet import task, reactor
import time

def timer():
    print time.ctime()

loop = task.LoopingCall(timer)
loop.start(1, now=True)
reactor.run()
}}}

{{{
foobar ~ # sh -c "sleep 5; date -s '07/02/2009 10:53:50'" & python testTimer.py
[1] 15635
Thu Jul  2 10:54:47 2009
Thu Jul  2 10:54:48 2009
Thu Jul  2 10:54:49 2009
Thu Jul  2 10:54:50 2009
Thu Jul  2 10:54:51 2009
Thu Jul  2 10:53:50 SAST 2009


^C[1]+  Done                    sh -c "sleep 5; date -s '07/02/2009 10:53:50'"
}}}

If the LoopingCall is left until ntpd updates the clock, it starts executing again. 

More info here: http://www.ehow.com/how_2083676_do-time-warp-dance.html

----------
Type     : defect
Component: core
Keywords : 
Priority : high
Nosy     : 
----------
http://twistedmatrix.com/trac/ticket/3899