[nanomsg] Re: Sorry for a stupid question. Is nanomsg about to see websocket transport?
- From: Samuel Tracol <samuel.tracol@xxxxxxxxxxx>
- To: nanomsg@xxxxxxxxxxxxx
- Date: Mon, 14 Apr 2014 19:11:33 +0200
Hello,
I have not yet posted anything, but I'll catch the opportunity to answer
to this post by also sharing some global feedback regarding nanomsg.
First of all, I would like the thanks Martin and all the author of this
great project : it really rocks ! It provides a very reliable solution
for network patterns.
We have a data-sharing system, event-driven, for our embedded system
(real-time room & speaker optimization), with nanomsg as first transport
implementation. It uses several schemes (PULL/PUSH, PUB/SUB, REQ/REP) to
implement the functionnalities needed.
A websocket back-end has been done : I choose Poco c++ as foundation lib
(http server and websocket server is already available, along with
several application utilities). This tiny webserver act as websocket
gateway to forward messages from the differents nanomsg enpoints into a
unique websocket connexion. With the help of javascript ArrayBuffer and
DataView, it has been possible to implement quickly a javascript client
part.
I'm maybe out of the subject, as a full javascript nanomsg client
implementation was not really a need for our project and a tunneled
solution was fine, but in few words :
- Dealing with multiples ports looks less elegant than URL path, in http
style.
- Merging all endpoints (all nanomsg sockets) into one tunnel is a
choice, more coherent in our specific application. It also allows to
have the exact same C code in all sides of the application : this help
to reduce the size of the javascript part.
- Having the nanomsg part inside the server allows to handle gracefully
the unexpected connection drop of the websocket from a higher level
application point of view.
Kind Regards,
Samuel
On 04/13/2014 07:03 AM, Garrett D'Amore wrote:
In a word. Close. As soon as I get some cycles. I want to do TLS first.
There is one unresolved question about websocket. That is how do we handle
paths in the server and port numbers seperately. There is no standard here. I
have some thoughts but I would appreciate feedback from folks who are deploying
websocket solutions.
Sent from my iPhone
On Apr 12, 2014, at 9:19 PM, crocket<crockabiscuit@xxxxxxxxx> wrote:
It seems there have been intermittent discussions about websocket
transport in nanomsg mailing list.
While people can just make nanomsg-websocket bridges, it'd be a lot
easier for users if nanomsg had websocket transport built-in.
How far is nanomsg from getting websocket transport along the way?
--
Samuel Tracol
Trinnov Audio
samuel.tracol@xxxxxxxxxxx / http://www.trinnov.com
5, rue Edmond Michelet 93360 Neuilly-Plaisance, France
Tel: +33 (0)1 47 06 61 37
Other related posts:
