[nanomsg] Re: Compressing TCP streams

  • From: Martin Sustrik <sustrik@xxxxxxxxxx>
  • To: nanomsg@xxxxxxxxxxxxx
  • Date: Mon, 02 Jun 2014 16:52:29 +0200

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I definitely think that websocket transport would be extremely useful,
whether it is implemented inside mangos, inside nanomsg or as a mangos
bridge to nanomsg.

Martin

On 02/06/14 16:12, Garrett D'Amore wrote:
> I still would like to do web socket.  I started the work on
> openssl, but to be honest nanomsg's internals are not terribly
> conducive to adapting to openssl.  There is a lot of work that
> relies on async io and fie descriptors, which makes it kind of a
> poor fit for adaptation to OpenSSL.
> 
> Writing the OpenSSL transport for mangos was trivial, by
> comparison. I suspect that websocket will be almost as trivial.
> 
> One of my ideas that I simply haven't done yet is to write a
> "device" command line application, that can be used so that you can
> use mangos as a trivial front end (device) to nanomsg, allowing for
> quickly deploying OpenSSL (and in the future websocket) on top of
> existing nanomsg applications without any need to recompile or
> change code.
> 
> Would folks find such an application useful?
> 
> My own immediate needs for mangos have subsided for a bit, as I'm 
> working on some other things right now, but if something like this
> would be useful I'd be happy to spend a little time to do both the
> device app, and websocket.  Just need to hear "votes" for it, is
> all, as I don't want to burn cycles on something nobody is going to
> use.
> 
> 
> On Mon, Jun 2, 2014 at 12:30 AM, Martin Sustrik
> <sustrik@xxxxxxxxxx <mailto:sustrik@xxxxxxxxxx>> wrote:
> 
> On 02/06/14 09:27, Martin Sustrik wrote:
> 
>> 2. Hop-by-hop compression. (This is what you mean AFAIU.) In
>> this case there may be a common context, but the compression must
>> be done within nanomsg and is, I reckon, protocol-specific (i.e. 
>> compression for reliable transports is going to be different
>> from compression for unreliable transports etc.) If you are
>> interested in this I would suggest creating a new transport,
>> something like "TCP with compression".
> 
> Btw, I still believe something like websocket transport may be a
> good idea. I also recall seeing some work on compressed websockets
> done in IETF. I don't know whether it have amounted to anything in
> the end though.
> 
> Martin
> 
> 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTjI+tAAoJENTpVjxCNN9YSQQIAKbv6AKGCgnZKMch8ecLeF3l
dA+tvxjHkrRMXPAz5xxRzDU/wRKG+o6WT0wcuviTuNMKCyeGOVAUSnwbja7ta+RX
PHEvZkiTtVOVExixTIgWMbZ9lKj2B4TGpP9+j9L1F6xvmu96XyqPJNAniL+FQ+q2
93c+nmzBbxfX5semghk0gnWF5W86y4ICJruTe+mW098shSFabPyx96o7dHNmM8hN
0tGaA+NEQUwY7784aJr/Ob69Mz1UTZICZeUibMLfjjIVJpHQKn6z6e8wwZmqkW8B
2wEDiMacuYfn6r55LeuOPP7SmqUUoHcxLupjaTUar9iLxzYZc8txUcodEfnn2ms=
=qoEv
-----END PGP SIGNATURE-----

Other related posts: