On Fri, Apr 12, 2013 at 2:48 AM, Gunnar Hellstrom <[email protected]> wrote: > - on the sender side I limit the maximum random sequence restart number to > 2**30.
Your recommendation of restarting the number is NOT as safe as randomizing. It increases the chances for collisions (temporary moments of corruption). Same reason why it's not as recommended to restart seq values to 0 anymore, even though it would work fine for the vast majority of use cases. Consider the use case of simultaneous logins. It's easier to do section "Keeping Real-Time Text Synchronized" if the seq values of all conflicting senders are dramatically different. Randomization ensures that the dramatic differences in seq values happens. That's for the use case if they don't keep track of separate real-time messages per full JID, and no, I'm not going to enforce the "full JID" requirement, since it complicates several implementations far more than the simple action of randomizing seq. Mark Rejhon
