[11:45:56] --- gww has joined
[11:56:52] --- SharonChisholm has joined
[12:01:21] <SharonChisholm> http://ops.ietf.org/lists/netconf.
[12:01:33] <SharonChisholm> Slides will be available there later
[12:01:37] <SharonChisholm> 4 drafts
[12:02:17] <SharonChisholm> titles of drafts have changed to be more aligned
[12:02:19] --- Yoshifumi Atarashi has joined
[12:02:24] <SharonChisholm> agenda bashing
[12:02:48] --- ray has joined
[12:02:51] <SharonChisholm> Andy - open issue on mailing list is about default operation ...
[12:03:07] <SharonChisholm> Andy - talked to Rob and Phil offline and they don't think it is a bug
[12:03:26] <SharonChisholm> Andy - if you want to replace the entire configuration .....
[12:03:39] --- dinakar has joined
[12:03:40] <SharonChisholm> Andy - if you don't want to nuke things, don't use a default of replace
[12:04:04] --- suz has joined
[12:04:05] <SharonChisholm> ------------------------
[12:04:16] <SharonChisholm> Rob - Netconf Protocol
[12:04:22] <SharonChisholm> Summary of chnages in -04
[12:04:33] <SharonChisholm> - details of subtree filtering
[12:04:46] <SharonChisholm> - added type attribute on filter elements
[12:05:01] <SharonChisholm> - Add #xpath capability
[12:05:13] <SharonChisholm> - update security considerations
[12:05:55] <SharonChisholm> - add default-operation paramater to <edit-config>
[12:06:29] <SharonChisholm> - <target> is now mandatory parameter for <lock> and <unlock>. There is noe default target
[12:06:42] <SharonChisholm> - remove XML schema for Netconf state Data
[12:06:50] <SharonChisholm> - remove data model stuff
[12:07:01] <SharonChisholm> - correct namespace handling a number of examples
[12:07:09] <SharonChisholm> - remove xml usage guidelines ...
[12:07:14] <SharonChisholm> - update reforat protocol xsd
[12:07:36] <SharonChisholm> Kept the 'no DTDs' guideline
[12:07:46] <SharonChisholm> added a typo
[12:07:51] <SharonChisholm> - fixed some typo
[12:08:07] <SharonChisholm> ------------------------------
[12:08:13] <SharonChisholm> Simon - please read the drafts
[12:08:29] <SharonChisholm> Simon - the biggest change is the section on filtering
[12:08:42] <SharonChisholm> Simon - errors left in as exercise for reader ;-)
[12:08:56] <SharonChisholm> Andy - we are in working group last call on all 4 documents
[12:09:17] <SharonChisholm> <November 28th>
[12:09:40] <SharonChisholm> Andy - not a lot on mailing list. Either address problems or people stopped reading. hopefully the latter
[12:09:46] <SharonChisholm> Andy - any other issues?
[12:09:56] <SharonChisholm> Dave P - so, you really want me to read it this time
[12:10:12] <SharonChisholm> Adny - that is why we threatlend you with working group last call
[12:10:45] <SharonChisholm> Jeff c - Positive comments before hand, negative send after the election
[12:11:14] <SharonChisholm> Andy - let's talk about the netconf state model. Chicken and egg thing. Can't put ina model without a modeling language
[12:11:36] <SharonChisholm> Andy - onj the other hand, we ned to be able to retreive this. operationally, we have a bit of a glitch
[12:11:51] <SharonChisholm> Wes - there are a number of points where you just point to future work, why no do that here?
[12:11:59] <SharonChisholm> Adny - not sure what the other 6 or 7 are ....
[12:12:12] <SharonChisholm> Andy - we talk about the operations
[12:12:21] --- yushunwa has joined
[12:12:24] <SharonChisholm> Wes - there are other things
[12:12:34] <SharonChisholm> Andy - add things that say this is future work.
[12:12:53] --- becarpenter has joined
[12:12:53] <SharonChisholm> Andy - gives some urgency to have some initial work to have the data model to make the protocol usuable
[12:13:12] <SharonChisholm> Glenn - not an objection to leaving it in ... one snippet.
[12:13:43] <SharonChisholm> Andy - what could change would be the representation of the schem. Could agree on the instance information.
[12:13:55] --- bert has joined
[12:13:58] <SharonChisholm> Sharon - the other thing would be to do the semantics, not the suntax
[12:14:16] <SharonChisholm> Rob - put it in and clean it up.
[12:14:32] <SharonChisholm> Andy - big warning that the representation is subject to change
[12:14:41] <SharonChisholm> Wes - put in as an example or it is concrete
[12:15:17] <SharonChisholm> Simon - let's put in this minimal grain of data model and say it is mandatory for the protocl work, but does not proclude future data model to be done differently
[12:15:28] <SharonChisholm> andy - could put it in a non-normative appendix
[12:15:44] <SharonChisholm> andy - someone comes up with something more complete later, this becomes deprecated later
[12:16:25] <SharonChisholm> wes - you are defining in the document how you get a session id ... not necessarily a bad thing, but ... you are defining today how you want it to work ... you either cycle at proposed or you are left iwth it forever
[12:16:43] <SharonChisholm> Andy - you can say you get the info without saying how.
[12:17:10] <SharonChisholm> Wes - that is what I was saying before ... word it carefullly, there is not standardization way to retrieve this, you might use example A in the meantime
[12:18:08] <SharonChisholm> Rob (other) - this is set for standards track ... anythin that gets put in there becomes a must. As per discussion in other area s.. things that ought to be mandatory. If you want to extend later on ... other things come along ... things in standards track need to be considered ... if not ready for that, should be there
[12:18:33] <SharonChisholm> andy - there are other bits of the data model like exchanging capabilities does not need to be therre. Not critical for any operations
[12:18:50] <SharonChisholm> Adny - could refuce the data model down to what is absolutely neede
[12:19:22] --- suz has left
[12:19:34] <SharonChisholm> Rob - by putting it in a particular namespace .. make it really small ... when real data model ... switch to different namespace .. then if people want to switch or support both for backwards compat, does not seem to seem bad
[12:19:50] <SharonChisholm> adny - hum -> want minimal data model in there; no hum; dont'
[12:20:41] <SharonChisholm> andy - show of hands ...
[12:21:12] <SharonChisholm> eliot - some to none hands in the room
[12:21:18] <SharonChisholm> <some to one>
[12:21:27] <SharonChisholm> Andy - other issues related to the documents?
[12:21:47] <SharonChisholm> Andy - random topics? Netmod?
[12:21:59] <SharonChisholm> Andy - some have looked at relax ng .. impressed
[12:22:20] <SharonChisholm> Glenn - relax ng ... is it something we should be considering for drafts here.
[12:22:34] <SharonChisholm> simon - it would not hurt to specifiy our little bits of schem in both.
[12:22:55] <SharonChisholm> andy - we have xsd
[12:23:11] <SharonChisholm> glenn - there have been stuff on netmod mailing list sayying they think it is neat (relax ng)
[12:23:19] <SharonChisholm> andy - which one is normative
[12:23:28] <SharonChisholm> glenn - would just have one
[12:23:41] <SharonChisholm> andy - perhaps a companion document
[12:24:13] <SharonChisholm> Juregen - started the relaxng discussion. I think leaving it is fine. the protocol is being read by protocol developers as appose to the data model stuff which has a much wider audience
[12:24:17] <SharonChisholm> andy - agree
[12:24:29] <SharonChisholm> andy - any other commnets on data modeling
[12:24:43] <SharonChisholm> andy - how do people feel there should be working group ... when ... and the scope
[12:25:16] <SharonChisholm> brian c - this not the only body that is having this discussion. Hard to answer without knowing what they do
[12:25:31] <SharonChisholm> brian c - if we do stuff, and others do stuff and vendors have to do both ... it would be bad
[12:25:40] <SharonChisholm> brian c - this is something we should be worried about
[12:26:05] <SharonChisholm> andy - my opinion is there is always going to be proprietary and standard. Leading and trailing edge
[12:26:32] <SharonChisholm> Brian c - I'm worried about more than one standard, not more than one proprietary model. very likely to happen given the current course and speak
[12:26:35] <SharonChisholm> <spead>
[12:26:59] <SharonChisholm> Adny - different communities that dont's want to view all the data in the same way. XMLified CLI
[12:27:18] <SharonChisholm> Andy - we ahve heard that from customers. we have also heard CIM. No homogenous view from the world.
[12:27:28] <SharonChisholm> Andy - I don't think there will ever be one model that makes people happy
[12:27:40] <SharonChisholm> Wes - there might be a valid case for multiple
[12:28:34] <SharonChisholm> wes - the problem comes that if you want interop. ... need to decide what your target audience is. If you target operators with one type of device . don't have to do much. if you want to talk to operators with multi-types ...
[12:29:09] <SharonChisholm> wes - to promote interoperability is to configure stuff in the same way. If we ignore ... either others will do it first or we end up with a not so usable protocll
[12:29:28] <SharonChisholm> andy - bottom up; start with framework stuff
[12:29:37] <SharonChisholm> wes - my point, any forward direction
[12:29:48] <SharonChisholm> Andy - even that limited charter would be harder than people think
[12:30:04] --- ray has left: Replaced by new connection
[12:30:17] <SharonChisholm> Andy - in the examples posted, willfully lacking in syntax ... I don't know what any of that stuff means .. until I get description clauses and other things ..
[12:30:21] --- ray_atarashi has joined
[12:30:26] <SharonChisholm> andy - most work in MIBs was probably in the description clause
[12:30:55] <SharonChisholm> andy - that is the real reason that protocol groups don't want to do configuration ...
[12:31:07] <SharonChisholm> glenn - it would be good to see who else is useful to see what the 'non usual suspects' think
[12:31:31] <SharonChisholm> Bert - are we turning this session into an informal bof (point of order)
[12:31:43] <SharonChisholm> bert - if so, close working group and have informal bof
[12:32:05] <SharonChisholm> dave h - question on recharter for this work .. parallel discussion for new group
[12:32:14] <SharonChisholm> Dave h - isn't it within scope to discuss this?
[12:32:27] <SharonChisholm> simon - good point ...
[12:32:50] <SharonChisholm> bert - kind of ok ... but such a topic needed to be on the agenda ... if dicuss ... do it informally.
[12:33:06] --- gorryf has joined
[12:33:08] <SharonChisholm> bert - not on the agenda for small things, but rechartering the working group is too big
[12:33:26] <SharonChisholm> andy - don't have a problem with closing the meeting and leaving it out of the miniutes. still interested in what people think
[12:33:53] <SharonChisholm> elliot - everything we do is confirmed by mailing lists. even if it informal here ... can be formal here
[12:33:58] <SharonChisholm> bert - informal bof
[12:34:12] <SharonChisholm> dave h - close this and ask you to leave ;-)
[12:34:34] <SharonChisholm> andy - more interested in what people want to do rather than where they want to do it.
[12:35:03] <SharonChisholm> bert - my first question was .. spending working group time ... don't want to see protoocl stuff come up next week that we should have talked here
[12:35:17] <SharonChisholm> andy - last time .. any issues with 4 documnets .... they want to discss
[12:35:37] <SharonChisholm> wes - wanted to reread the document ... lock object ought to apply to stuff in a configuration store as well
[12:35:40] <SharonChisholm> elliot -= wht
[12:36:01] <SharonChisholm> wes - copy config end up copying the state of things ... no warning it is incomplete ...
[12:36:37] <SharonChisholm> wes - copy candidate to running ... someone half through editing candidate
[12:36:40] <SharonChisholm> andy - lock the other one
[12:36:49] <SharonChisholm> wes - still working on arguemnts
[12:37:07] <SharonChisholm> wes - better to lock candidate so you can't copy it
[12:37:25] <SharonChisholm> andy - fi I lock candidate and I lock running, then I'm pretty sure they will be both stable
[12:37:57] <SharonChisholm> wes - this is how you ought to use locks ... should be used carefully and condiser all paths you are going
[12:38:17] <SharonChisholm> andy - text will be in the copy config ... stable operation ... lockng both ... ensure no change ...
[12:38:22] <SharonChisholm> etc
[12:38:53] <SharonChisholm> wes - not the copy config that bothers me
[12:39:10] <SharonChisholm> wes - dummer people would lock, edit then lock
[12:39:36] <SharonChisholm> wes - design transaction protocol to allow minor tweak . no notion that someone else is part way through
[12:40:27] <SharonChisholm> phil - trading that against their ability to get the configuration at all. if someone has it locked, you can't have it at all
[12:40:50] <SharonChisholm> andy - cli today, how important is this issue
[12:41:06] <SharonChisholm> bert it seems to be that at least you want to write some text about these issues
[12:41:10] <SharonChisholm> andy - agree
[12:41:39] <SharonChisholm> andy - lock all db you will be dealing with, work then unlock them all. Seems intuitive. Maybe it isn't
[12:41:52] <SharonChisholm> Bert - capture phil's stuff too
[12:42:43] <SharonChisholm> wes - interesting thing about ietf is we do stuff backwards . we produce these documents ... write the mib and publish stuff and think that people will figure it all. They often don't we do BCP. We should do best future practice
[12:42:56] <SharonChisholm> andy - i agree with you. the disagree i have with you is normative should versus must
[12:43:15] <SharonChisholm> andy - you should lock a database. to say your must never be able to ... that goes to far
[12:43:25] <SharonChisholm> wes - i was not trying to imply should must ...
[12:43:31] <SharonChisholm> andy - you did last time
[12:43:35] <SharonChisholm> wes - that was security
[12:43:53] <SharonChisholm> elliot - moving on :S
[12:44:27] <SharonChisholm> simon - we do stuff for interoperability .. you also say we should specify operations ..
[12:44:39] <SharonChisholm> simon - separation between protocol and mechanisms
[12:45:26] <SharonChisholm> andy - agree with wes ... mibs in the past .. rational for a lot of decisions does not make it into drafts. itwould make it easier to understand . 'not part of protocol ... ripped it' not always good
[12:45:58] <SharonChisholm> simon - doing BCP on these things. You can think if excellent thoughts on problems that may occur when people use it ... think that you should write a book on this problem instead
[12:46:25] <SharonChisholm> simon - some people get rick on this sort of thing (%)
[12:46:55] <SharonChisholm> wes - the exmaples in the document are good
[12:47:11] <SharonChisholm> andy - agree. make people aware of the operational considerations section in the document
[12:47:42] <SharonChisholm> andy - any other issues
[12:48:19] <SharonChisholm> andy - working group meeting is officially closed ...
[12:48:24] <SharonChisholm> andy - unoffical meeting/
[12:48:41] <SharonChisholm> andy - posting examples .... interseted in seeing relax ng versus xml schema.
[12:48:56] <SharonChisholm> andy - read the drafts
[12:49:30] <SharonChisholm> andy - documents have iana considerations
[12:49:38] <SharonChisholm> eliot - netconf over beep
[12:49:44] --- toro_toro has joined
[12:49:49] <SharonChisholm> eliot - i think i can do it ... there is a webpage
[12:49:57] <SharonChisholm> bert - yeah, web page
[12:50:04] <SharonChisholm> andy - ok ....
[12:51:50] <SharonChisholm> done
[12:51:53] --- SharonChisholm has left
[12:53:40] --- Yoshifumi Atarashi has left
[12:54:51] --- toro_toro has left
[12:59:37] --- gorryf has left
[13:00:32] --- dinakar has left: Replaced by new connection
[13:03:18] --- becarpenter has left
[13:04:33] --- dinakar has joined
[13:04:44] --- dinakar has left
[13:08:08] --- yushunwa has left: Disconnected
[13:09:14] --- ray_atarashi has left: Disconnected
[13:12:38] --- dinakar has joined
[13:12:59] --- dinakar has left
[14:09:34] --- bert has left: Disconnected
[14:11:41] --- gww has left: Disconnected
[14:29:18] --- ray_atarashi has joined
[14:29:46] --- ray_atarashi has left
[14:46:08] --- bert has joined
[14:46:22] --- bert has left