SCTP item - RE: [cgl_discussion] POC Cancelled

Venkata Jagana jagana at
Tue Nov 11 08:56:38 PST 2003

Glad to hear this perspective.


                      "Wilson, Andrew"                                                                                                         
                      <andrew.wilson at>         To:       Venkata Jagana/Beaverton/IBM at IBMUS, <sdake at>                       
                      Sent by:                          cc:       "Fleischer, Julie N" <julie.n.fleischer at>, cgl_discussion at  
                      cgl_discussion-bounces at lis        Subject:  RE: SCTP item - RE: [cgl_discussion] POC Cancelled                           
                      11/11/2003 08:20 AM                                                                                                      

I don't think steering has any philosophical objections to
tweaking the 2.0 requirements so they continue to track the



-----Original Message-----
From: cgl_discussion-bounces at
[mailto:cgl_discussion-bounces at] On Behalf Of Venkata
Sent: Monday, November 10, 2003 4:36 PM
To: sdake at
Cc: Fleischer, Julie N; cgl_discussion at
Subject: SCTP item - RE: [cgl_discussion] POC Cancelled

This is a very valid add-on requirement and this *must* be
added to the list right away but there is no other way since the
corresponding function in RFC 2960 is already depracated.

The reasoning for inclusion of this RFC is as follows and this
should be taken forward to the specs group:

There already exists a checksum algo in RFC 2960 and which was
determined to be fundamentally weak and so, the IETF came up with a
replacement checksum algo described in RFC 3309.

According to the text from RFC 3309:

"This document updates and replaces the
 Adler-32 checksum definition in [RFC 2960].
 Note that there is no graceful transition mechanism
 for migrating to the new checksum. Implementations
 are expected to immediately switch to the new
 algorithm; use of the old algorithm is deprecated."

Our linux implementation already switched to the algo described
in RFC 3309 as mandated and as I mentioned already, this is already
in 2.6 kernel. If we don't add it in the reqts list, basically
the CGL reqt would be behind but otherwise, distros would pick up
the new code since old code doesn't exist anymore. In addition,
our linux implementation would be interoperable with other
non-linux implementations.


                      Steven Dake

                      <sdake at>                To:
"Fleischer, Julie N" <julie.n.fleischer at>

                      Sent by:                          cc:
cgl_discussion at

                      cgl_discussion-bounces at lis        Subject:  RE:
[cgl_discussion] POC Cancelled


                      11/10/2003 04:06 PM

                      Please respond to sdake


I will add them to the errata list and we can review on the mailing list
before submitting to specs.  I'm a little concerned about the last
requirement and won't submit to specs without a more thorough
explination about the reasoning for the addition.  I suspect specs would
be concerned about adding requirements without a useful reason...


On Mon, 2003-11-10 at 16:26, Fleischer, Julie N wrote:
> > Because of a team meeting in my company, I will be unable to host
> > 11/12/03 POC meeting.  I will publish the errata shortly, and that
> > errata can be reviewed in the specs mailing list.
> Steve -
> Not sure if these can make it in without POC discussion, but, at this
week's POC, I was going to bring up the items we mentioned briefly at
end of last POC for the errata list:
> Per 11/4/03 email by Julie Fleischer:
> >Found a minor bug on page 50. "For example, AVL.7.1 Process-level
> >Non-intrusive...." should be changed to "For example, AVL.8.1
> >Process-level..." The AVL.7.1 should be changed to AVL.8.1...
> >
> >-Terence
> Per 11/4/03 email by Venkat Jagana:
> >The SCTP requirement (STD 4.0) lists RFC 1112
> >(Host extensions for IP multicasting)and we believe that
> >it's a mistake - It isn't an SCTP supporting
> >RFC.
> >So, please remove RFC 1112 from the list there.
> >The SCTP maintainer suggests that we should add
> >RFC 3309 (SCTP Checksum Change) to the SCTP
> >RFC list. This RFC is already supported within
> >the code which is part of 2.6 kernel.
> Thanks.
> - Julie
> **These opinions are not necessarily those of my employer,**

cgl_discussion mailing list
cgl_discussion at

cgl_discussion mailing list
cgl_discussion at

cgl_discussion mailing list
cgl_discussion at

More information about the cgl_discussion mailing list