[21:48:28] --- ScottH has joined
[21:49:14] --- ScottH has left: Replaced by new connection
[21:49:14] --- ScottH has joined
[21:49:36] --- hardie has joined
[21:49:57] --- pigdog has joined
[21:57:16] --- sleinen has joined
[22:02:12] --- rpe has joined
[22:03:01] --- bert has joined
[22:06:28] --- Ted Faber has joined
[22:08:07] --- sommerfeld has joined
[22:08:12] <hardie> no jabber scribe, eh?
[22:09:24] --- Ted Faber has left
[22:11:53] <sleinen> Sorry Ted, we don't seem to have found a Jabber scribe.
[22:12:14] <sleinen> Sharon is taking notes, so at least we'll have a good transcript afterwards...
[22:12:16] <hardie> no worries; thanks for letting me know
[22:12:45] <pigdog> sharon: starting data model discussion
[22:13:02] <pigdog> sharon: met on monday, people agreed on the scope of the discussion
[22:13:24] <pigdog> sharon: looking at both abstract and concrete information models
[22:13:42] <pigdog> sharon: start with physical interfaces and bgp as case studies
[22:14:16] <pigdog> sharon: next steps, continue discussion, draft charter, and we would like to have a bof in san diego
[22:14:45] <pigdog> sharon: what's the minimal data modeling we need to get the protocol to work?
[22:14:54] <pigdog> sharon: that stuff needs to be in the base protocol spec
[22:15:23] <pigdog> sharon: supporting SMI for NETCONF would not be necessary to use NETCONF protocol
[22:15:34] <pigdog> sharon: this has ramifications
[22:15:53] <pigdog> sharon: we want more structure in the error codes, and some of this needs to be in the base and some will need to be in the data model
[22:16:01] --- bkhabs has joined
[22:16:05] <pigdog> sharon: comments from andy on that one?
[22:16:16] <pigdog> andy: still need to flush out details on this one.
[22:16:28] <pigdog> sharon: need to revisit section 7 of the protocol draft
[22:16:39] <pigdog> sharon: some of that stuff should move to the SMI for netconf
[22:16:55] <pigdog> andy: thank you sharon
[22:17:03] <pigdog> andy: please join the mailing list
[22:17:16] <pigdog> andy; very important issues; not pure syntax
[22:17:22] <pigdog> andy: there's more to this.
[22:18:02] <pigdog> rob enns, protocol draft
[22:18:09] <pigdog> -02 changes
[22:18:17] <pigdog> rob: several features removed
[22:18:35] <pigdog> rob <rpc-progess> <rpc-abort> <rpc-abrt-reply> removed
[22:18:44] <pigdog> rob: channels and notifications went as well
[22:19:07] <pigdog> rob: fixed the uri format; put version at the end
[22:19:10] <pigdog> rob: fixed getconfig
[22:19:19] <pigdog> rob: removed lock capability from appendix b
[22:19:40] <pigdog> <get-all> becomes <get>
[22:19:53] <pigdog> <get-state> becomes <get-all>
[22:20:00] <pigdog> CORRECTION
[22:20:10] <pigdog> it's all becoming <get>
[22:20:41] <pigdog> q: why did you remove notifications?
[22:20:58] <pigdog> rob: asychnronouos notifications were difficult to implement in some substrates
[22:21:41] <pigdog> andy; low demand, possible subject for future work
[22:22:13] <pigdog> q: {can't understand}
[22:22:42] <pigdog> q: need notification data model
[22:23:12] <pigdog> andy; notifications REQUIRE channels
[22:23:22] --- bkhabs has left
[22:24:12] --- sob has joined
[22:24:41] <pigdog> andy; more comments about difficulty of implementing channels in the base
[22:24:55] <pigdog> andy: particular issues with single stream approach
[22:25:01] <pigdog> chris lonvick:
[22:25:46] <pigdog> chris: speaking for the syslog wg, we're looking at making some base decisions on the syslog protocol that will spur revisions of RFC 3195. If NETCONF doesn't want to deal with it we might make this an issue for our group.
[22:25:59] <pigdog> randy preshun
[22:26:23] <pigdog> randy: both speakers raised data model issues. requirements in this area should be brought to the attention of the data model list
[22:26:57] --- memo56 has joined
[22:27:25] <pigdog> andy: netconf/soap discussion
[22:27:30] --- memo56 has left
[22:27:32] --- rei has joined
[22:27:44] <pigdog> andy: "must understand" must not be ignored...
[22:27:54] <pigdog> andy: some issues relating to HTTP as a substrate
[22:27:59] <hardie> What's the context of that statement?
[22:28:09] <hardie> Oh, the "use of HTTP as a substrate BCP"?
[22:28:44] <rpe> yes
[22:28:57] --- sakai has joined
[22:29:20] <pigdog> sorry- went to the mike
[22:29:51] <sleinen> I put diffs between -00 and -01 of draft-ietf-netconf-soap-... on http://www.ops.ietf.org/netconf/draft-ietf-netconf-soap-00-01-diff.html
[22:32:06] <pigdog> andy: issues
[22:32:13] <pigdog> 1.5: validation conformance
[22:32:47] <pigdog> andy: beyond scope of what referential integrity checks might be performed
[22:32:53] <pigdog> andy: need a data model for that
[22:33:35] <pigdog> randy; minor comment: current text might prohibit referential integrity
[22:33:42] <pigdog> andy: maybe take the text out
[22:33:55] <pigdog> randy: let's just be sure not to leave ambigous language in
[22:34:19] <pigdog> andy: syntax check = xsd validation?
[22:34:55] --- leslie has joined
[22:35:15] <pigdog> 5.1 end of message directive
[22:35:54] <pigdog> andy: consensus seems to be that we should have something that is not legal CDATA or we end up with a shell escape
[22:36:00] <pigdog> proposed consensus
[22:36:24] <pigdog> "]]>]]"
[22:36:41] <pigdog> andy: any comments?
[22:36:44] --- leslie has left
[22:36:57] <rpe> ]]>]]>
[22:37:09] <pigdog> oops thx
[22:37:20] <pigdog> 5.2 SSH port assignment
[22:37:35] <pigdog> andy: consensus matches what the operators were telling eliot
[22:37:42] <rpe> I was wondering if the xmpp implementation would escape it properly :-)
[22:37:57] <pigdog> andy; we should tell vendors in using some SHOULDs
[22:38:05] <pigdog> andy: comments?
[22:38:41] <pigdog> 7.1 list of criteria for mandatory transport
[22:39:21] <pigdog> andy; if we're supposedly doing this for operators, we should do SSH
[22:39:59] <pigdog> margaret: if the operators prefer ssh they'll get that plus what vendors want (assuming a difference)
[22:40:34] <pigdog> SHOULD support BEEP; MAY support SOAP
[22:41:24] <hardie> there is no MUST?
[22:42:13] --- Milkshake has joined
[22:42:13] --- Milkshake has left
[22:42:37] <ScottH> MUST support SSH
[22:43:07] <hardie> Thanks for the clarification.
[22:44:56] <pigdog> eliot: need to separate agents and managers
[22:46:05] <pigdog> {eliot of debate}
[22:54:42] --- sob has left: Disconnected
[22:54:55] <pigdog> 8.1 <get-all>
[22:56:21] <pigdog> tim stoddard: no filter capability
[22:56:33] <pigdog> tim: it's a dump
[22:56:58] <pigdog> andy: you can specify the start point in the data model and you'll get instances and subtrees that are appropriate
[22:57:46] <pigdog> original get-all was get-state
[22:57:59] <pigdog> andy: we had this discussion in the interim
[22:58:14] <pigdog> andy: need to separate state from config
[22:58:19] <pigdog> andy: that was from the operator
[22:58:41] <pigdog> andy: it put a requirement on the vendor to do the separation
[23:00:13] <pigdog> tim: verify that get-state was an incorrect requirement
[23:00:16] <pigdog> andy: yes
[23:00:52] <pigdog> tim: <get-all> interfaces would return many thousands of elements
[23:09:03] --- hardie has left
[23:12:50] --- Ted Faber has joined
[23:17:07] --- Ted Faber has left
[23:18:43] --- sakai has left
[23:28:25] --- pigdog has left
[23:29:33] --- ScottH has left
[23:30:32] --- bert has left: Logged out
[23:43:41] --- rpe has left
[23:51:01] --- rei has left