Just my 2 cents: get/set are easy to remember, while peek/poke are... kind of weird
On Thu, Feb 16, 2012 at 12:03 AM, Andrew Hume <[email protected]>wrote: > works for me. > > On Feb 15, 2012, at 3:57 PM, Pieter Hintjens wrote: > > Hi all, > > I'd like to canvas opinion for some cleanup of the ctx creation API. > This would be a backwards compatible change in accordance with > http://www.zeromq.org/docs:policies#toc2. > > There are several faults with the current context API: > > * It is inconsistent with the rest of the API, e.g. zmq_init for a new > context vs. zmq_msg_init vs. zmq_socket for a new socket. > * It still has that iothreads option in zmq_init, which IMO should > have gone a long time ago. > * It is not extensible, so we are seeing new experiments like > zmq_init_thread_safe(). > > My proposal is to make a new consistent API for contexts: > > * zmq_ctx_init and zmq_ctx_term > * zmq_ctx_peek and zmq_ctx_poke to get/set options > * explicit option settings for ZMQ_IOTHREADS and ZMQ_THREADSAFE > > And internally this would create the context lazily, only when the > first socket is created (so the IOTHREADS can be applied then). > > There is a separate discussion about changing the functions for > get/set message options. Peek/poke are short. getctxopt/setctxopt are > messy wrt the other ctx functions. > > -Pieter > _______________________________________________ > zeromq-dev mailing list > [email protected] > http://lists.zeromq.org/mailman/listinfo/zeromq-dev > > > > ------------------ > Andrew Hume (best -> Telework) +1 623-551-2845 > [email protected] (Work) +1 973-236-2014 > AT&T Labs - Research; member of USENIX and LOPSA > > > > > > _______________________________________________ > zeromq-dev mailing list > [email protected] > http://lists.zeromq.org/mailman/listinfo/zeromq-dev > > -- Aurélien Vallée +33 6 47 41 70 37
_______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
