Steven McCoy wrote: > The additional complexities in the real world are different delays for > different exchanges, and frequently a delay spool larger than the system > memory, requiring disk offload. > > Often you will also see conflation to reduce the number of updates too, > especially when clients may be Internet based.
Exactly. Maybe it would be worth of making the delayer a new type of device (zmq_delayer) and refining it a little. Btw, the disk offload will arrive naturaly once the disk-offload feature is ported from 0MQ/1.0 (next month hopefully). Martin _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
