Re: [rtcweb] Comments on draft-miniero-rtcweb-http-fallback

[Salvatore Loreto <[e-mail]>]

On 10/18/12 8:47 PM, Martin Thomson wrote:
> On 18 October 2012 10:30, Lorenzo Miniero <[e-mail]> wrote:
>> http://www.ietf.org/mail-archive/web/rtcweb/current/msg05035.html
> I remember reading that, but considered that starting a new thread was
> appropriate, that's all.
>
> My view on the subject is that firewall circumvention is not and
> should not be the goal of any such work.  A proper approach would be
> to use a TURN server that manages access control in a transparent
> fashion.
>
> That said, firewalls often block UDP for reasons other than a deep and
> abiding hatred of VoIP.  Those firewalls are probably not doing
> anything that would prevent something like WebSockets from working.
> In those environments, a standard method for using the available
> communication channels could be helpful.  WebSockets is also far more
> efficient with its use of bits.  Hence, WebSockets.
>
> The main objection to something like WebSockets is that some
> environments actively block WebSockets.  These firewalls often block
> HTTPS in the same way.
that is interesting,
I have heard a lot people saying that usually the firewalls
(well some time they talk of the proxy really more then firewall)
are more likely to block WebSocket then HTTP

and they use that statement to promote an HTTP fallback
(here I am talking in general not specifically for webrtc)

However if we are talking of secure WebSocket and https fallback
I agree that if a firewall block secure Websocket then it will most 
likely block
HTTP as well !

>   In that case, I see no reason to continue to
> build ever more elaborate circumvention methods.
then I tend to agree in not trying to define a fallback of the fallback
>
> I've said enough.  I just realized that this is not really an rtcweb
> topic.  (behave?)
I would say that is rtcweb until we do not define what we want to do

ciao
Salvatore

-- 
Salvatore Loreto, PhD
www.sloreto.com