On Fri, Nov 28, 2014 at 3:24 AM, Cameron Simpson <c...@zip.com.au> wrote: > > With multiprocessing they're completely separate (distinct memory spaces); > data > must be passed from one to the other, and there's a cost for that.
Single-machine IPC can use shared memory, not just message passing via pipes and sockets. multiprocessing works readily with ctypes arrays and numpy arrays. It's just that using shared memory doesn't scale to distributed systems. Tighter coupling isn't generally worth the tradeoff in scalability, resiliency, and security. It has its uses in terms of client/server architecture on a single machine. Windows NT has always made extensive use of shared memory for communicating between subsystem servers and clients (e.g. csrss, lsa, and conhost) using its LPC protocol. _______________________________________________ Tutor maillist - Tutor@python.org To unsubscribe or change subscription options: https://mail.python.org/mailman/listinfo/tutor
