W3C home > Mailing lists > Public > www-http-ng-comments@w3.org > July to September 1998

SMUX protocol

From: by way of Henrik Frystyk Nielsen <leijon@ludd.luth.se>
Date: Sun, 02 Aug 1998 18:52:44 -0400
Message-Id: <3.0.5.32.19980802185244.009c5930@localhost>
To: www-http-ng-comments@w3.org
First of all I presume there is a typing error in the latest version of
the draft. It concerns the subsection titled "Protocol ID's". The sentence
reads "Protocol ID's above 0xfffff are atoms." Considering what is said
earlier in the paragraph should it not read 'above 0x1ffff'?? And on a
related note, why limit yourself to 256 atoms where you could easily have
made 64K of them without any problems that I could see?

Secondly I have a suggestion, this concerns Control Messages. The function
DefineEndpoint is said to be useful aong other things for firewall
proxies. I agree, but should it not be useful to define a control message
that was UnDefineEndpoint, since for a proxy it could be useful to
undefine certain areas. For example proxy.foo.com might be willing to
give connections to all of http:// but it might want to say that
internal.foo.com is not reachable through the proxy.  Might  be useful for
some implementations of the smux. 

Perhaps even EndpointQuery could be a good control message, to ask
explicitly if a certain resource is available or not. But I am not as
certain in this case, unless for certain kinds of proxies where it would
be unwieldy to list the resources accesible with DefineEndpoint
statements.

Yours, 
  Viktor Leijon 
  (leijon@ludd.luth.se) 
Received on Sunday, 2 August 1998 18:52:22 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 24 September 2003 13:07:28 EDT