Very good, thanks! Does it possible to get this working on udp protocol? 2014-03-24 5:10 GMT+04:00 Garrett D'Amore <garrett@xxxxxxxxxx>: > All, > > I’ve finished some debugging, and am now ready to announce the first public > release of my SP protocol implementation in pure Go. > > https://bitbucket.org/gdamore/sp > > There’s also documentation here: https://godoc.org/bitbucket.org/gdamore/sp > > This is licensed under Apache License 2.0 > > This is *VERY* preliminary. In particular, I’ve only implemented REP, REQ, > XREP, XREQ over TCP and IPC at this point. (I’ve not tested interop between > nanomsg with IPC yet, but it seems to work with TCP, and I’d be somewhat > surprised if it doesn’t just work with IPC.) The implementation of each of > those protocols is only about 100 lines each. The implementation of IPC and > TCP are also only about 100 lines each. (Admittedly, they make use of a > common core that is a little bit larger, but I expect TLS to only require > about 100-150 lines. Including whitespace and comments. :-) > > I will probably implement Pub / Sub in the next day or so. I expect few if > any challenges there. I’ll have to provide a trie, and I was somewhat > surprised to see that nanomsg handles the subscription filters on the > subscriber side rather than the publisher. (I guess this makes much less > work for the publisher to keep track of things, and a vastly simpler > protocol, but it also assumes network bandwidth is cheap. It would work > better over a true broadcast/multicast medium too.) > > The IPC also only supports Unix like operating systems. golang has no native > support for named pipes, and I don’t have a Microsoft platform to test any > implementation on, even if I were tempted to write one myself (and to be > honest, I’m not. :-) > > I would appreciate any *useful* feedback anyone might have. The bitbucket > repo has a tracker. For the moment the issue tracker accepts anonymous > posts. (Until such time as I get spam and have to disable anonymous issue > posting. :-) > > I’d also welcome any collaboration, if folks here want to contribute. > > - Garrett > > -- Vasiliy Tolstov, e-mail: v.tolstov@xxxxxxxxx jabber: vase@xxxxxxxxx