[zeromq-dev] ZeroMQ WebSocket Transport

Michael Haberler mail17 at mah.priv.at
Tue Aug 12 16:45:46 CEST 2014


Bjorn,

Am 12.08.2014 um 15:50 schrieb Bjorn Reese <breese at mail1.stofanet.dk>:

> On 08/12/2014 01:58 PM, Doron Somech wrote:
> 
>> ZWS is a ZeroMQ over WebSockets, the protocol is raw and subject to
>> changes. After some discussion with Michael Haberler we agreed that
>> having a binary protocol is better (currently the protocol support text
>> messages only), however the RFC is not yet updated. You can read the rfc
>> at http://rfc.zeromq.org/spec:39
> 
> I have only looked very briefly at this, so I may be missing something.
> 
> Have you considered using the WebSocket continuation opcode instead of
> encoding the more flag into ZWS? That way all ZMTP meta-data can be
> handled outside the WebSocket payload, and the latter will be an exact
> copy of the ZMTP body.

reading RFC6455... yes there is in fact such a thing, Continuation Frames ;)

I am admittedly your bona-fide JS/ws idiot; that said, I still have to discover a client-side JS Websockets API for that feature, nevermind one which covers the majority of browsers, but I might be overlooking something?

dont you think that we'd be pushing the envelope a bit by relying on this feature?

I rather go for "Websockets is a dumb frame pipe, not even binary-safe until proven otherwise" as a bottom line assumption

- Michael

ps: it'd be great if we could move this discussion on https://github.com/zeromq/JSMQ/issues - focused and low background noise 

> 
> _______________________________________________
> zeromq-dev mailing list
> zeromq-dev at lists.zeromq.org
> http://lists.zeromq.org/mailman/listinfo/zeromq-dev




More information about the zeromq-dev mailing list