RE: TCP Mapping

["Liu, Hong" <Hong.Liu@neustar.biz>]

I would like to stay focused on the issue I raised in this thread. If I was
not clear in previous emails, here we go again:

Issue: what would be the proper way to tag server-pushed messages from the
server?

Option 1: Use one bit in the Total Length header in the EPP datagram. This
is specific to the TCP transport mapping.

Option 2: Use <extension> child element of <epp>. This is transport neutral.

There may be other options, but those two are what I've seen so far.

I initially suggested option 1, and Scott has suggested option 2. I still
don't know what Eric's opinion is.

I want to hear what others on the list think.

Cheers,

--Hong

-----Original Message-----
From: Eric Brunner-Williams in Portland Maine
[mailto:brunner@nic-naa.net]
Sent: Saturday, June 29, 2002 8:31 AM
To: Liu, Hong
Cc: 'ietf-provreg@cafax.se'; brunner@nic-naa.net
Subject: Re: TCP Mapping 


> That was my understanding. The issue is of general interest, and I would
> like to get more feedback on the list.

Which issue?

[snip...]
> So Eric, you are for Scott's proposal to use the <extension> child element
> for defining the PUSH.

I don't think I've seen it.

[snip...]

That said, the selection of tcp as transport has no relation to where
state is held, or which end of a stream was the stream's initiator.

These are (possibly gratuitous) design choices.

Eric