On 22/08/13 15:09, Jan Dvořák wrote:
On Thu, 22 Aug 2013 14:27:39 +0200 Martin Sustrik <sustrik@xxxxxxxxxx> wrote:For starters I would propose following limits: - max message size - max number of peers - max total size of subscriptions Anything else you can think of?Maximum number of subscriptions *per peer*, or otherwise it's wide open to a trivial denial of service attack.
Right. That's on the publisher side.
Otherwise these should suffice to signal users that security is a concern. By the way, XREP will be included, right?
nanomsg's equivalent to XREP is "nn_socket (AF_SP_RAW, NN_REP);" Martin