Saving state against reboot

Marcus G. Daniels marcus@cathcart.sysc.pdx.edu
13 May 1997 06:15:42 -0700


>>>>> "PP" == Paul Prescod <papresco@calum.csclub.uwaterloo.ca> writes:

PP> What if LispOS checkpoints every five second (on average). What if
PP> I have a mail queue that stores outgoing messages. Then I hit
PP> "send". A message goes out. The machine crashes before that action
PP> is checkpointed. I reboot the computer and it starts the thread
PP> where it was last checkpointed and sends the message again. This
PP> says to me that applications must control checkpointing using
PP> (dumplisp) or (save-object ...)

In my scheme of grouped transactions, any time important coordination
with the outside world was going to occur the grouping parameter could
be temporarily set to 0 (or some small number).  That way, the fact a
message was sent would be recorded.