[10:58:43] --- leslie-ietf has joined
[11:42:42] --- guenther has joined
[11:49:36] --- j.schoenwaelder@jabber.eecs.jacobs-university.de has joined
[11:50:44] --- Barry Leiba has joined
[11:58:24] --- hta has joined
[11:59:10] --- leslie-ietf has left
[12:01:54] --- j.schoenwaelder@jabber.eecs.jacobs-university.de has left
[12:04:22] * guenther starts scribing
[12:04:30] <Barry Leiba> Yay, P!
[12:04:42] * guenther growls
[12:05:01] <guenther> agenda bashing...
[12:05:24] <guenther> (Is anyone here not physically in the apparea meeting?)
[12:06:29] --- dcrocker has joined
[12:06:46] <guenther> added: IBE (identify based encryption) brought up in relation to email architecture
[12:07:07] --- presnick has joined
[12:07:22] <guenther> announcements:
[12:07:33] <guenther> apps area workshop in Feb
[12:07:43] <guenther> 10-11? 11-12?
[12:07:49] <guenther> requires position paper
[12:07:58] <guenther> deadline for those is the 14th
[12:08:37] <guenther> - a full day of interop testing at next IETF?
[12:08:58] --- leslie-ietf has joined
[12:09:37] <guenther> - NFSv4 protocol has hit issues with UTF-8 vs POSIX/SUS
[12:09:47] <guenther> looking for volunteers to help them out with that?
[12:10:38] <guenther> (from room) affects file:// scheme?
[12:10:59] <guenther> Patrick Fahlstrom volunteered
[12:11:12] <guenther> - IESG looking for note takers
[12:11:14] --- frodek has joined
[12:11:32] <guenther> (for the IESG meetings narrative minutes)
[12:11:34] --- Lisa has joined
[12:11:39] <guenther> next: IBE auth
[12:12:09] <guenther> - came up during IESG review of the IBE crypto stuff
[12:12:16] --- presnick has left
[12:12:25] <guenther> - draft-ietf-smime-ibearch-06.txt
[12:13:07] <guenther> - email/directory protocols support SASL, so adding this would be easy for them
[12:13:26] <guenther> - security protocols for certs however, are based on HTTP, which doesn't have SASL
[12:13:51] --- resnick has joined
[12:14:07] --- resnick has left
[12:14:07] --- resnick has joined
[12:14:15] <guenther> alexey: time for webauth WG? there will be bar BOF tomorrow; meet at 6pm at message board
[12:14:55] <guenther> V-card dinner tomorrow, 8pmish @message board
[12:15:09] <guenther> next: BCP 56
[12:15:35] <guenther> (Mark Nottingham)
[12:15:36] --- bernard.desruisseaux has joined
[12:16:08] <guenther> is BCP 56 still BCP?
[12:16:09] --- sm-msk has joined
[12:16:09] --- tonyhansen has joined
[12:16:36] <guenther> <<"Substantially new services" should not reuse existing ports.>>
[12:16:43] --- rohan has joined
[12:16:51] <guenther> Is this still felt to be right?
[12:17:17] <guenther> similarly <<new protocols or services should not reuse http: or other URL schemes>>
[12:17:36] --- randy has joined
[12:17:37] <guenther> other minor technical quibbles
[12:18:07] <guenther> main point is that people are using http for machine-readable data not intended for browsers
[12:18:16] --- kmurchison has joined
[12:18:23] <guenther> some services are explicitly designed to leverage HTTP's features
[12:18:43] <guenther> "extending vs tunnelling"
[12:18:54] <guenther> (latter bad, former okay?)
[12:19:06] --- Eric Allman has joined
[12:19:21] <Barry Leiba> What is the RFC number for BCP 56?
[12:19:39] <guenther> RFC 3205
[12:21:14] <guenther> Larry Masinter: BCP was written as direct warning to WGs (IPP)
[12:21:45] <guenther> randal gellens: number of things driving that document then that still apply today, document has not had that large of impact
[12:22:02] <guenther> rg: still see use of http/port 80 because that goes through firewalls
[12:22:17] --- klensin has joined
[12:22:23] <guenther> rg: problem hasn't gone away, solution still 'works'
[12:22:48] <guenther> rg: http is easy to whip up too
[12:22:57] <guenther> patrick: was area director at time
[12:23:30] <guenther> pf: was not intent to say "don't use HTTP", but rather to give things to think about before making choice
[12:23:50] <guenther> pf: some items may no longer be relevant and other issues may be missing now
[12:24:46] <guenther> lisa: other areas using http to do RPC (cert fetching, etc)
[12:25:14] <guenther> lisa: you can call it a webservice, however, the whole ecosystem of http is ignored
[12:25:18] --- yangwoo has joined
[12:26:02] <guenther> lisa: e.g., they don't say whether servers must support HTTP features like conditional gets, resulting in servers that don't support it
[12:26:27] --- atarashi has joined
[12:27:00] <guenther> lisa: to address randy's point, perhaps we should have a profile of http, or list of http features that each service should consider
[12:27:36] --- apparea has joined
[12:27:42] --- apparea has left
[12:27:46] <guenther> mark: http normally fails gracefully; is you doin't support cond-get, you send whole response, which is okay
[12:27:57] --- ray has joined
[12:28:07] <guenther> dave crocker: mostly in agreement with lisa
[12:28:48] <Lisa> RFC 2616: " The semantics of the GET method change to a "conditional GET" if the request message includes an If-Modified-Since, If-Unmodified-Since, If-Match, If-None-Match, or If-Range header field. A conditional GET method requests that the entity be transferred only under the circumstances described by the conditional header field(s)."
[12:28:52] <guenther> dc: use of http as 'transport' is common and often minimal
[12:29:09] --- sm-msk has left: Logged out
[12:29:19] <guenther> dc: a survey doc may been good thing
[12:29:23] --- dromasca has joined
[12:29:26] <guenther> mark: disagree violently
[12:29:27] --- bert has joined
[12:29:53] <guenther> mark: world has been moving to using http more fully
[12:30:19] <guenther> mark: not every app uses every feature of http, but that's okay, browsers don't even do that
[12:30:35] <Lisa> hhmm, I'm not finding a great quote, but other than "If-Match", there are conditionals that do not fail gracefully if the server ignores them.
[12:31:29] <Lisa> " A cache or origin server receiving a conditional request, other than a full-body GET request, MUST use the strong comparison function to evaluate the condition."
[12:31:34] <guenther> mark: profiling as shutting doors on future enhancements?
[12:32:10] --- dwd has joined
[12:33:23] <randy> I agree with Mark's last comment, of clarifying and strengthening BCP 56 so that if HTTP is being used for its features it is more clear what that means and how to do it, but if HTTP is being used as a transport, it says "Don't do this"
[12:33:29] <Barry Leiba> What was the result of that discussion?
[12:33:39] <Barry Leiba> Will someone work on updating BCP 56?
[12:33:46] <Barry Leiba> Or do we think it's fine as it is?
[12:33:54] <Lisa> Action item: "Volunteer to work with Lisa and Mark"
[12:34:01] <guenther> mark: think we should produce clarification document more fully elucidating the tunnelling vs _using_ choice and
[12:34:06] <guenther> ...
[12:34:12] <randy> I can help out a bit
[12:34:31] <guenther> can someone summarize Mark's last statement more completely?
[12:34:40] <Barry Leiba> I'm pleased to help, but don't have a lot of practical experience with random HTTP usage.
[12:34:49] * guenther needs more coffee
[12:34:57] <klensin> ...(that agreement was from me... I think we need to be _very_ clear about boundaries and what is and is not appropriate. And that probably requires significantly reopening the document'
[12:35:11] <Lisa> Philip do you take it with milk, sugar, other?
[12:35:22] <guenther> milk, no sugar
[12:35:40] <resnick> What a great chair!
[12:35:43] <guenther> next: SIP file transfer
[12:36:20] <guenther> whoops
[12:36:30] <guenther> next: file metadata
[12:36:41] <guenther> originated from SIP file transfer work
[12:37:12] <guenther> see drafty-garcia-sipping-* for background, why SIP, uses cases
[12:37:22] --- sodabrew has joined
[12:37:41] <guenther> idea: file-metadata XML document:
[12:37:50] <resnick> Basically, he wants an XML hunk of data to describe a file.
[12:38:10] <randy> (John, I was agreeing with Mark's final comment independently of your agreement)
[12:38:18] --- kdz has joined
[12:38:24] <guenther> - identity, immuatable properties (size, URN, MIME type)
[12:39:09] <guenther> - instances describe metadata associated with a particular 'endpoint' (URI, mod-date, read-date, etc)
[12:39:52] <guenther> file-metadata doc can also express partial changes, using XML patch operations framework
[12:40:07] <guenther> can be used with periodic notifications, etc
[12:40:20] <guenther> questions:
[12:40:40] <guenther> - do we (IETF) need to define a general format like this?
[12:40:56] <guenther> - if so, how to deal with other, similarly purposed formats?
[12:41:01] <guenther> -- webdav properties
[12:41:27] <guenther> -- METS (Metadata Encoding and Transmission Standard) (from library of congress?)
[12:41:33] <guenther> -- MIME
[12:41:35] <guenther> -- others?
[12:42:11] <guenther> (for MIME: content-disposition, etc)
[12:42:18] <guenther> continue the work?
[12:42:42] <guenther> Pete Resnick: perfectly reasonable to hone in on our own standard
[12:42:59] <guenther> pr: if METS works for you, then documenting that would be good
[12:43:15] <guenther> pr: sounds like MIME will not be itself
[12:43:22] <guenther> s/itself/enough/
[12:43:32] <guenther> pr: profile another format?
[12:43:55] <guenther> john klensin: agrees with Pete
[12:44:13] <randy> Was there a draft for a MIME type for directories a few years ago?
[12:44:17] <guenther> jk: warning: don't go on deep excursion into file description
[12:44:28] <guenther> jk: simplicity is our friend; complexity is our enemy
[12:44:43] <guenther> jk: a few simple things with clear scope would be good
[12:45:19] <guenther> larry: metadata is hard; simple terms prove to have different meanings
[12:45:51] <guenther> lm: saying less can be the Right Thing
[12:46:47] <guenther> rohan mahy: you've mentioned METS, has it been delved into fully?
[12:46:56] <guenther> rm: written out a sample on whiteboard?
[12:46:58] <guenther> no
[12:47:35] <guenther> randy: few years ago, was involved in ISO filetransfer
[12:47:39] <hta> should we have a support group called OSIers Anonymous?
[12:47:43] <resnick> OSI
[12:48:03] <guenther> rg: "no meaning has different meaning for different people"
[12:48:07] <guenther> [sic]
[12:48:31] <guenther> rg: the less you transfer, the better the interop
[12:48:48] <guenther> lisa: simplicity is motherhood/applepie
[12:49:10] <guenther> ld: this would still be a useful building block
[12:49:43] <guenther> ld: example: MUAs using email to transfer stuff
[12:50:39] <randy> temptation with file xfer is to export file systems semantics
[12:50:42] <guenther> jk: there's a society that creates grand unified theories of expressing this stuff and falls completely into a big hole
[12:50:49] <randy> that doesn't interop very well
[12:51:13] <guenther> jk: must be simple to understand or will create more confusion
[12:51:52] <guenther> next: netconf modelling language
[12:52:24] <guenther> (Sharon Chisholm to present)
[12:52:50] <sodabrew> randy: the mime type for directories is rfc 2425
[12:53:55] <guenther> netconf needs a way to represent data models; this is a proposal that does so using XML schema
[12:53:57] --- julian.reschke has left
[12:54:10] <guenther> draft-chisholm-netconf-*
[12:54:21] <guenther> result of offline disscussions with OEMs
[12:54:30] <guenther> (slides are posted)
[12:55:04] <guenther> approach:similar to how we defined MIBs, minus some stuff, plus some other stuff
[12:55:23] <sodabrew> randy: oh wait, 2425 might be "directory" as in "directory services" (oops).
[12:55:27] <guenther> uses XML schema, plus defines appinfo tags to fill in gaps
[12:56:04] <guenther> (can use off the shelf tools, etc)
[12:56:26] <guenther> already XML on many netconf managed devices, etc
[12:57:08] --- kdz has left
[12:57:40] --- hta has left
[12:57:40] <guenther> (see slide for complete list of "areas of requirements")
[12:58:24] <guenther> proposal is known to be clumsy for relationships/key-references
[12:58:36] --- hta has joined
[12:58:36] <guenther> because XML schema is weak there
[13:00:13] <guenther> some modelling considerations too
[13:00:35] <guenther> point: modelling using XML schema is quite usable/workable
[13:01:35] <guenther> lisa: forgot mention: think this would be interesting for configuring application servers in future
[13:02:00] <guenther> next: the YANG counter-proposal
[13:02:05] --- kdz has joined
[13:02:11] <guenther> why we need this (a new work)
[13:02:25] <guenther> netconf base in RFC 4741
[13:02:41] <guenther> defines RPC mech and ops
[13:02:55] <guenther> leaves content and data models for future work == now
[13:03:35] <guenther> currently, most config is via CLI (cisco's CLI mostly)
[13:03:53] <guenther> we have verbs now, but we lack nouns
[13:04:38] <guenther> rohan: two words about data modelling vs information modelling?
[13:05:13] <guenther> netconf currently is like SNMP without SMI (the thing we define MIBs in)
[13:05:52] <guenther> goal: interoperable netconf models
[13:06:25] <guenther> so, multivendor design group cam up with YANG
[13:06:31] --- hardie@jabber.qualcomm.com has joined
[13:06:36] <guenther> they think it's fairly well baked
[13:06:53] <guenther> large, but lots of examples
[13:07:47] <guenther> have tools to generate XSD and RelaxNG from YANG for manager apps
[13:08:27] <guenther> (list of netconf-specific issues, similar to what sharon described)
[13:09:08] <guenther> YANG uses XPATH for naming instances
[13:09:22] <guenther> augmentation based addition similar to SNMP
[13:10:04] <guenther> Phil Schaffer [sic?] presents now
[13:10:10] <guenther> why not XML schema?
[13:10:45] <rohan> note: augmentation in yang can be selective based on an XPath expression called 'where'
[13:10:46] <guenther> constraints on the finished product, commit time enforced, not on the specific changes on the wire
[13:11:08] <guenther> operations guided by the model: semantics drive the syntax
[13:11:49] <guenther> simple XML validation is not sufficient: 5 scenarios where that isn't true. e.g., mandatory fields not present in deletes
[13:11:55] <guenther> (thanks rohan)
[13:12:37] <guenther> XSD suffers from extended subset syndrome: can't directly leverage knowledge of other communite
[13:12:42] <guenther> community, even
[13:13:26] <guenther> what are you left with if you use XSD: - tools can't enforce constaints by themselves - knowing the schema language isn't sufficient
[13:13:44] <guenther> similar to what happened with SMI and ASN.1
[13:13:56] <guenther> SMI does not match current ASN.1!
[13:14:31] <guenther> YANG aimed to meet the exact requirements
[13:14:58] <guenther> based on three proprietary DMLs and their years of experience
[13:15:15] <guenther> draft is based on running code, usable in current form
[13:16:42] <guenther> discussion before YANG found that all the vendors invented their own thing(s) after each found their users couldn't hack stuff based on general work
[13:17:36] <guenther> Leif: what about OWL? think you have reinvented it
[13:17:45] <guenther> peter: have not looked at OWL
[13:18:24] <guenther> leif: think the semantics of YANG are similar to OWL
[13:18:47] <guenther> ?: think use of schema languages were dismissed too easily
[13:19:14] <guenther> ?: think there may be buried problems in YANG that wouldn't be discovered for a few years
[13:19:34] <resnick> Point of order question: Do we need to decide this in here? Isn't this a question for NETCONF?
[13:19:46] <guenther> mark nottingham: agree, doesn't buy dismissal of schema languages
[13:20:06] <Lisa> Pete: We don't decide anything here; we need to figure out some way of getting the sense of this population (suggestions on possible hums welcome)
[13:20:58] <guenther> mn: two points: 1) versioning is _very_ hand; 2) data binding are only implicit in schema, but everyone does them
[13:21:10] <resnick> Well, I guess I should refine my question: On what are we getting a sense of the population? A sense of whether these things should go forward? A sense of whether Apps should care about this? ....?
[13:21:43] <guenther> mn: OWL may be a good place to start
[13:22:28] <guenther> andy bierman: (netconf cochair) have been struggling with XSD for years
[13:23:03] <guenther> ab: only 7% of group in meeting thought they understood it
[13:23:13] <guenther> ab: relaxng didn't excite them
[13:23:43] <guenther> ab: main goal: having something easy enough that people can develope in it
[13:24:13] <guenther> ab: relaxng hasn't stopped the flow of schema languages
[13:24:38] <guenther> ab: need stable data models, think netconf-specific language is best path there
[13:25:17] <guenther> rohan mahy: think andy's comments assume that they will get it right first time
[13:25:26] <guenther> ab: SMI choose stability over new features
[13:26:10] <guenther> rm: comparing sharon's draft vs YANG: 1) extensibility in YANG is very ambitious, but sharon's doesn't have enough
[13:26:52] <guenther> 3) constraints are very strong in YANG
[13:27:17] <guenther> 2) just modeling actual config objects...? (missed)
[13:27:52] <guenther> ? from ericcson: this is not the first try: SMI, SMIng, something else, this
[13:28:04] <guenther> main audience is operators, not developers
[13:28:39] <guenther> think sharon's draft is only about half what it would need
[13:29:16] <guenther> ops areas director: "the IESG never makes mistakes, only the previous one did"
[13:29:18] --- trond has joined
[13:29:46] <guenther> need data model to make netconf work in the real world
[13:30:35] <rohan> rohan: 2) might be better to model just the running-configuration semantics and syntax, than everything you can express in netvconf
[13:30:40] <guenther> reiterate: the users of netconf are not programmers
[13:30:48] <guenther> must speak their language
[13:30:53] --- frodek has left: Replaced by new connection
[13:30:56] <guenther> (rohan thanks again!)
[13:31:51] <guenther> leslie daigle: not finding XSD experts in the netconf WG is no surprise
[13:31:58] <guenther> apps area is where this expertise is
[13:32:13] <guenther> e.g., crisp WG and IRIS
[13:33:17] <guenther> ld: would like discussion to continue in the apps area, with specific bits from netconf
[13:33:39] <guenther> guidelines on using schema
[13:33:39] <hardie@jabber.qualcomm.com> Note that a lot of IRIS folks have switched to relaxNG, at least in my experience. They still have xsd expertise, of course.
[13:34:09] <guenther> leif: not specific to netconf: looking at writing data model for kerberos servers
[13:35:28] <guenther> sharon: at start, relaxng wasn't ready, but that has grown; if you do netconf-specific language, you get netconf-specific knowledge
[13:35:37] <rohan> i am also interested in being able to use this for sip device configuration using the SIP UA Profile framework: draft-ietf-sipping-config-framework-14
[13:35:48] <guenther> larry: coauthored BCP on XML schema languages
[13:36:07] <guenther> lm: not volunteering to update that doc, but perhaps it should be
[13:36:39] <guenther> ted: should your point be repeated at mic?
[13:36:55] <hardie@jabber.qualcomm.com> sounds like things have moved on; no need.
[13:37:17] <guenther> andy: if only 7% of the people in a working group understand the data modelling language, you won't get mibs
[13:37:46] <hardie@jabber.qualcomm.com> (I'm actually across the hall; I would have come over for this, but I agreed to scribe MEDIACTRL, so I'm following it from here).
[13:37:47] --- hta has left
[13:37:48] <guenther> rm: not the intent to just point people at the XSD spec
[13:38:12] <guenther> ted: wow, my brain would explode if I tried to scribe and follow elsewhere
[13:38:48] <guenther> ab: we need 95% understand the language; XSD never go that uptake
[13:38:53] --- amarine has joined
[13:39:14] <guenther> ?: how many people understand yang? (i.e.. any more than xsd?)
[13:39:32] <guenther> ab: not many yet, but it better match should make it easy
[13:40:13] <guenther> david: hope is that yang is simple enough that you can just look at the stuff and read it
[13:40:16] --- dwd has left
[13:40:17] <resnick> Hubris such a lovely feature.
[13:40:28] --- dwd has joined
[13:40:32] <kdz> Balazs Lengel speaking
[13:40:42] <guenther> can you summarize what he said?
[13:40:54] <guenther> mark: yes, you understand it because you wrote it
[13:41:03] <kdz> No, I was focusing on his badge not his comments
[13:41:42] <guenther> mark: re: schema: there seems to be an implicit use of XML that isn't really a requirement
[13:41:54] <Lisa> you need an ontology for configuration data regardless of whether you have a data format, or a data modeling language.
[13:42:01] <guenther> dave crocker: no clue which is better, but sounds familiar at meta level
[13:42:39] <guenther> dc: debate is buried in details, not looking at higher question
[13:43:01] <guenther> dc: distinction of syntactic vs semantic modelling is fundamental
[13:43:20] <klensin> Lisa, This all sounds like an argument , based on the "I have a favorite hammer, everything looks like a nail" principle, being made by people with different favorite hammers
[13:43:40] <guenther> dc: XML is useful syntactic generalization but semantic web shows that it doesn't above that
[13:43:56] <Lisa> John: if so, I'm not sure how to uplevel or resolve the hammer battle.
[13:44:16] <Barry Leiba> I agree with John.
[13:44:17] <guenther> kdz, can you scribe for a bit?
[13:44:27] <klensin> One might (shudder) consider doing some "architecture" :-(
[13:44:51] <guenther> rm: andy argument that stuff must be simple is tautological
[13:45:01] <Lisa> It seems very difficult to separate ontology-building requirements from modeling language requirements.
[13:45:36] <guenther> rm: "will the real requirements please stand up?"
[13:47:18] <guenther> ops AD: long background on this (CMIP, SNMP, workshops...)
[13:48:09] <guenther> AD: multiple companies working together, much effort put into YANG, to make netconf work in real world
[13:49:37] <guenther> ab: think application area isn't showing convergence on this
[13:50:17] <kdz> Ray Atarashi speaking
[13:50:55] <guenther> ?: netconf is good interface, but we need higher level model, maybe using OWL/RDF?
[13:51:06] <guenther> ra: need uses cases
[13:51:32] <guenther> mark: think effort is great
[13:51:57] --- sm-msk has joined
[13:51:57] <guenther> mn: but, already use relaxng, with tools in many languages
[13:52:35] <guenther> ab: have yang->XSD translater
[13:52:51] <guenther> mn: need the eco-system of tools
[13:53:26] <guenther> david harrington: RFC 3535, working on network management, talked about what we need for config purposes
[13:53:54] <guenther> dh: before you try to solve the problem, understand the problem space, read that for some bits
[13:54:06] <guenther> scott bradner: what are we talking about?
[13:54:25] <guenther> sb: yang folks asked for bof and were turned down? why?
[13:54:46] <guenther> lisa: some procedural reason (missing drafts)
[13:55:16] <guenther> lisa: who would attend with expertise?
[13:55:29] --- hta has joined
[13:55:35] <guenther> sb: it's really necessary, but why not in a bof
[13:56:28] <guenther> ops AD: draft of yang was in time, the justification draft didn't exist
[13:57:13] <guenther> chris newman: wanted to have the "why new work instead of existing <modelling language>" instead of dragging that into the bog
[13:57:15] <guenther> bof
[13:57:23] <amarine> heh
[13:57:25] <guenther> sb: okay, but that should have been stated up front here
[13:57:32] <Lisa> freudian slip!
[13:57:35] <guenther> yeah
[13:57:48] * guenther points at his keyboard
[13:58:02] <Barry Leiba> Scott's very right about one thing, at least: it's a Bad Idea to turn the apparea meeting into a de-facto BOF when the real BOF has been denied.
[13:58:59] <guenther> john klensin: have been trying to follow discussion, both on list and here, and haven't been seeing much actual discussion
[13:59:05] <guenther> jk: just a comparison of hammers
[13:59:53] <hardie@jabber.qualcomm.com> I thought Rohan's "little rules" discussion was quite useful.
[13:59:57] <guenther> jk: not hearing an engaged discussion about how to move forward without having multiple tools all moving forward in competition
[14:00:07] <hardie@jabber.qualcomm.com> I'd love to see that analysis of some of what I've done in ECRIT.
[14:00:40] <guenther> leslie: didn't think andy's comment about applications being configured with random config files was useful
[14:00:41] <Barry Leiba> I have a "Braveheart"-like image, of a bunch of people on one side of a field with hammers of one kind, and a bunch of people on the other side of the field with hammers of the other kind... and someone screams, and everyone runs to the middle of the field and they start bashing each other with their respective hammers.
[14:01:00] <guenther> ld: app area has something to learn from netconf
[14:01:42] <guenther> ld: "yang is simple to read and understand" argument seems thin, as it is only 7months ago
[14:02:27] <guenther> ld: having discussion in netconf will be tricky: had XML discussion at start of netconf and didn't get the involvement then
[14:03:50] --- julian.reschke has joined
[14:04:54] --- michaelsinger has joined
[14:05:01] <Lisa> it is an interesting question of where the onus is...
[14:05:02] <guenther> netconf co-chair: why fear of DSL?
[14:05:33] <Lisa> Simon Leinen was at mike
[14:05:37] <guenther> thanks
[14:05:46] * guenther left out too much there
[14:05:58] <guenther> (losing traction...)
[14:06:01] --- ray has left
[14:06:15] --- ray has joined
[14:06:32] <guenther> dave crocker: many reasons for coming to the IETF: to get stuff bless and to get real engineering work
[14:06:55] <guenther> dc: the latter often gets the response "prove you aren't an asshole", thus proving that _we_ are
[14:07:34] <guenther> dc: equally, people coming to the IETF that respond to critique with "no no, it's simple" is the opposite problem
[14:07:43] <guenther> dc: neither is a dialoge
[14:07:57] <guenther> dc: this stuff is important, we need a real dialoge
[14:08:16] <guenther> larry: a config file is an application, reading it is an operation
[14:08:17] <guenther> ?
[14:08:59] <guenther> lida: what would be helpful hums?
[14:09:02] <guenther> lisa, even
[14:09:06] --- sm-msk has left
[14:09:26] <guenther> lisa: should we have bof on the netconf data modelling topic?
[14:10:02] <guenther> rohan: a bof around requirements for possible solutions for creating syntax and semantics for netconf and other protocols?
[14:10:57] <guenther> burt (I think): each time we get the requirements, in netconf we got the objects...
[14:11:08] <guenther> just delayed stuff for 1,2 years
[14:11:09] --- hardie@jabber.qualcomm.com has left: Disconnected.
[14:11:19] <guenther> andy: concerned with boiling the ocean
[14:12:18] <guenther> david: this is a netconf problem
[14:12:44] <guenther> david: we _have_ the requirements?
[14:12:48] --- Eric Allman has left
[14:12:51] <guenther> s/?//
[14:13:07] <guenther> bert: we don't want the 2 year discussion
[14:13:45] <guenther> mark: very circumspect about our ability to bless stuff in this solution space
[14:14:08] <guenther> mn: our ability is to give warnings
[14:14:58] <guenther> ops AD ( Dan Romascanu)
[14:14:58] --- fenton has joined
[14:16:19] <guenther> dan:believe yang should continue, perhaps with parallel tracks
[14:16:45] <guenther> dan: nice to have a full XSD draft so we can compare apples with applies
[14:16:48] <guenther> apples, even
[14:17:17] <guenther> dan: do not think we should stop to talk about requirements
[14:18:19] <guenther> barry leiba: dave said something like "we have a netconf problem that should be solved yesterday", are we going to get concensus here?
[14:19:36] <guenther> chris newman: "if we did a bof with yang documents and other docs (XSD based), do preople think we should do that?"
[14:19:57] <guenther> strong rough concensus in favor of the question
[14:20:11] --- dcrocker has left
[14:20:53] <guenther> open mic comments:
[14:21:32] <guenther> tony: 2821bis is in late call, please comment; discussion of implementation reports would be good too
[14:22:19] <guenther> dave crocker: I voted "no" because: it's a good topic, but groups get bogged down because they don't agree on what the goal is
[14:22:37] <guenther> dc: requirement docs help when they get people on the same page
[14:23:20] <guenther> chris: would having each proposal give a statement of goals would be a good thing?
[14:23:22] <guenther> dc: at least
[14:23:32] --- fenton has left
[14:24:10] <guenther> Ronald Bonica: bringing uses cases and example showing how the proposals apply would be goo
[14:24:11] <guenther> d
[14:24:49] --- dromasca has left
[14:25:02] <guenther> dave harrington: have time reserved to discuss a proposal (transforming SMI into XSD) in the ops area meeting
[14:25:11] <guenther> dh: people have pushed back
[14:25:25] --- j.schoenwaelder@jabber.eecs.jacobs-university.de has joined
[14:25:40] <guenther> dh: there's going to be a presentation of a comparison of the yang vs smi vs xsd approaches
[14:26:32] --- resnick has left: Lost connection
[14:26:43] --- bert has left: Logged out
[14:26:44] <guenther> dh: this is unseen (by dave), but you can see it tomorrow
[14:29:00] <guenther> lisa: <series questions aimed at non-netconf people that devolves>
[14:29:10] --- Lisa has left
[14:29:10] --- Barry Leiba has left
[14:29:15] --- atarashi has left
[14:29:23] --- kdz has left
[14:29:32] --- randy has left
[14:29:49] --- rohan has left
[14:29:52] <guenther> meeting ends
[14:29:57] --- hta has left
[14:30:43] --- ray has left: Computer went to sleep
[14:31:44] --- yangwoo has left: Computer went to sleep
[14:34:07] --- guenther has left
[14:36:50] --- sodabrew has left
[14:38:21] --- michaelsinger has left
[14:43:28] --- bernard.desruisseaux has left
[14:45:07] --- j.schoenwaelder@jabber.eecs.jacobs-university.de has left
[14:45:40] --- klensin has left: Disconnected
[14:46:00] --- leslie-ietf has left
[14:46:45] --- trond has left: Disconnected
[14:48:49] --- amarine has left
[14:52:34] --- dwd has left
[14:58:48] --- hta has joined
[15:19:10] --- hta has left
[15:21:07] --- kmurchison has left
[16:10:57] --- m_ersue has joined
[16:14:20] --- m_ersue has left
[16:52:00] --- bert has joined
[16:52:06] --- bert has left
[20:03:43] --- tonyhansen has left: Replaced by new connection