Generically, it can take any set of variables in the original output fact i, no, slot, i, num, a, t' and determine a new destination p.
In Paxos, since we partition on slot, it only needs the slot number slot and acceptor a.
The function itself will be a mapping from each location a to its set of partitions p, using slot mod n to find the right partition, where n is the number of partitions.
My partitioning transformation introduces the distribution function
slotPartition
below.Generically, it can take any set of variables in the original output fact
i, no, slot, i, num, a, t'
and determine a new destinationp
. In Paxos, since we partition on slot, it only needs the slot numberslot
and acceptora
.The function itself will be a mapping from each location
a
to its set of partitionsp
, usingslot mod n
to find the right partition, wheren
is the number of partitions.