To:
"'ietf-provreg@cafax.se'" <ietf-provreg@cafax.se>
From:
"Gould, James" <JGould@verisign.com>
Date:
Mon, 17 Dec 2001 10:39:42 -0500
Sender:
owner-ietf-provreg@cafax.se
Subject:
RE: More Private Comments
Scott, I have no problem with items 1 and 2, but item 3 "no "content length: " header in TCP transport.", I have a different proposal. If there is a desire to use a packet header, as an implementer I would find it more efficient and simpler to use a network header that only includes the length of the packet as a network byte order, four byte, unsigned integer. By using the network header, the packets will be framed and the framing can be done without having to do string comparisons. Jim > ---------- > From: Hollenbeck, Scott > Sent: Monday, December 17, 2001 7:47 AM > To: 'ietf-provreg@cafax.se' > Subject: More Private Comments > > Last week I received a few private comments that will result in changes to > the EPP documents. Here's a quick summary: > > Eric Brunner-Williams provided some suggested changes to the data > collection > policy elements. Nothing radical, so I intend to make the changes he > requested. > > Dan Manley also sent me some private comments touching on 4 things: > > 1) auth info not required for transfer query command > > This was discussed on the list a little while ago. I intend to change > things such that auth info is required to complete a transfer query. > > 2) inconsistent poll response format > > This comment described an issue with the way <poll> messages are returned. > In the current documents, the response <msg> element is overloaded to > return > the message. Dan suggested a change the to the 1301 response code so that > it has a consistent meaning, and putting the message within the <resData> > element. I disagreed with putting the message with <resData> because > <resData> is currently used only for _object specific_ response data, and > I > suggested a compromise of putting the message within the <msgQ> element, > like this: > > S: <response> > S: <result code="1301"> > S: <msg>Message dequeued.</msg> > S: </result> > S: <msgQ count="5"> > S: <qDate>2000-06-08T22:00:00.0Z</qDate> > S: <polledMsg id="12345">Transfer requested.</polledMsg> > S: </msgQ> > S: <resData> > S: <obj:trnData xmlns:obj="urn:ietf:params:xml:ns:obj" > S: xsi:schemaLocation="urn:ietf:params:xml:ns:obj obj.xsd"> > S: <obj:name>example</obj:name> > S: <obj:trStatus>pending</obj:trStatus> > S: <obj:reID>ClientX</obj:reID> > S: <obj:reDate>2000-06-08T22:00:00.0Z</obj:reDate> > S: <obj:acID>ClientY</obj:acID> > S: <obj:acDate>2000-06-13T22:00:00.0Z</obj:acDate> > S: <obj:exDate>2002-09-08T22:00:00.0Z</obj:exDate> > S: </obj:trnData> > S: </resData> > S: <trID> > S: <clTRID>ABC-12345</clTRID> > S: <svTRID>54321-XYZ</svTRID> > S: </trID> > S: </response> > S:</epp> > > This puts all of the <poll>-specific data in the <msgQ> element, keeps > <resData> object-specific, and makes the response code structure > consistent. > At the end of our last private exchange Dan said he still wasn't > comfortable > with this suggested modification, so we may have more to talk about. > > 3) no "content length: " header in TCP transport. > > Dan's comment is that it would be helpful for buffering purposes for the > application reading from a socket to know the size of an incoming PDU. he > and I talked about it a bit, and came with a suggested change to the TCP > mapping to define a short header of the form "type:length:value", where: > > Type = "application/epp+xml" > > Length: length of the value in octets, represented in hex. > > Value: The <epp>...</epp> stuff. So a TCP datagram containing an XML > instance of length 128 decimal (80 hex) would look like this: > > application/epp+xml:80:<epp>...</epp> > > 4) Domain and Contact transfers in EPP > > Dan and I talked a bit more about the comments he made during the SLC > provreg session, but no specific changes were agreed to. > > Eric and Dan, please feel free to expand on anything you think needs more > explaining or if you disagree with my summations. > > -Scott- > >