[04:31:57] ab joins the room [04:32:19] ab leaves the room [07:39:12] smc joins the room [07:39:18] smc leaves the room [07:53:44] sharon joins the room [07:56:36] brabson joins the room [08:00:31] andy joins the room [08:01:04] blue sheets [08:04:19] agenda - text file [and eye chart] [08:07:21] [08:08:30] ----- [08:08:33] NETMOD architecture [08:09:21] slide 2 [08:09:58] why isn't yin a single document? [08:10:42] slide 3 [08:11:57] slide 4 [08:12:08] "Where does yang fit?" [08:12:50] slide 5 [08:15:48] oops - we are on slide "Use Case #3" [08:16:19] "Use case #4" [08:18:01] send input at this point directly to phil and lada [08:18:33] ---------------- [08:18:37] Martin [08:18:45] on use case 2, how does the agent generate the proper elements with this approach [08:19:26] asked [08:19:51] Juergen Schoenwaelder joins the room [08:20:24] DavidAlanReid joins the room [08:20:44] b* - isn't that more a question for the DSDL mapping? [08:21:01] OK [08:21:08] ray joins the room [08:21:13] bclaise joins the room [08:21:23] ---------- [08:21:30] Martin (take 2) [08:22:05] Yang High-level Presentation; Yin; XML on the wire [08:22:29] slide 1 [08:22:31] sorry 2 [08:22:35] "Yang High-level" [08:25:20] "yang concepts" [08:25:56] [doesn't it get messy when you have your state and configuration data interleaved?] [08:26:13] "Yang module" [08:26:39] [you do not have to interleave if you do not want ;-] [08:26:50] [thanks ... very helpful ;-)] [08:27:01] [It makes more sense to interleave though] [08:27:21] "data definitions" [08:28:43] "strucure 1(3)" [08:28:48] slide 6 [08:29:07] no 'choice' on slide 5... [08:29:29] andy - is that editorial or do you want me to say it at the mike? [08:30:05] structure 2(3) [08:30:11] just editorial I guess -- it is an important construct [08:31:56] Q - is there a way with a list or something else to allow modeling a set? Where order is not important [08:31:59] A - that is the default [08:32:29] A - this says it can carry semantic meaning [08:32:36] dave h - example does not show ordering [08:32:43] A - right, have not shown everything [08:33:11] L* -an unorderedl list is not the same as a set; anything can only be there once [08:33:22] A - a list is a set [08:33:44] wes - are you going to go into when keys are unique? Can identify uniqeness via keys [08:33:51] a - keys have to be unique [08:34:07] a - might require multiple keys sometimes [08:34:15] slide 8 - structure 3(3) [08:35:00] slide 9 - reusable simple types [08:35:18] l* - back one slide ... not a set? [08:35:20] A - still a set [08:35:37] l* - can't be. otherwise 2 threes would not be allowed. [08:35:44] A - not allowed to have that. [08:35:59] l* - that would be strange. Or am I misunderstanding something? [08:36:12] l* - arrat of basic types [08:36:23] [08:36:44] you can have N duplicates -- there is no key in a leaf-list [08:38:26] dave h - take to list [08:38:36] slide 9 [08:38:53] edit with ordered-by user, insert attribute [08:39:24] andy -didn't parse that. Comment for the mike? [08:39:41] no, Dave says take it to the list [08:39:47] ok [08:39:58] slide 10 [08:40:17] slide 11 [08:40:56] slide 12 [08:42:42] end of yang; talk about yin [08:42:42] slide 13 [08:43:03] q - [08:44:57] l* - is the xml represenatation unique? or can there be multiple XML for a single yang [08:45:00] a - unique [08:45:20] a - except for the usual whitespace and namespace, etc [08:45:44] l* - need uniquness or you can't use XSLT [08:45:56] dave p - the rules will be in the document [08:46:10] slide 14 [08:46:58] slide 15 [08:47:34] slide 16 [08:47:45] dlpartain joins the room [08:47:54] [this is a lot more tutorial then I was expecting] [08:48:49] simon joins the room [08:49:30] tony - is the intention to just have a mapping from multiple modeules and sub modules into a single XML File? [08:49:32] a - no [08:49:43] slide 17 [08:50:00] bert joins the room [08:50:16] slide 18 [08:51:18] slide 9 [08:51:29] slide 20 [08:52:21] slide 21 "Free Yang Tools" [08:52:23] oops, both servers on port 25... [08:53:13] Q - in yang there is a feature caled instance identifier. the xml representation is xpath; another option [08:53:39] Q - why was xpath chosen since the other way does not require understanding xapth [08:53:40] FYI: yangdump now called yangtools, includes yangdump and yangdiff [08:53:53] a - like subtree? [08:54:00] Q - [08:54:45] Dave p - suggest post alternative to mailing list [08:55:25] l* - primary reason is xpath is actually a standard [08:55:34] l* - basic types in yang and mapped to xsd types [08:55:36] a - yes [08:55:52] l* - you have other definitions as well. Do you have xsd definitions for these? [08:56:01] a - that has more to do with DSDL mapping [08:56:20] l* another reason to make sure xml rep is unique [08:56:38] juergen - currently no mapping to xsd for base datatypes, but we can discuss [08:56:44] "Things to work on" slide 22 [08:59:00] -------- [08:59:07] Juergen [08:59:12] Common Yang Data Types [08:59:39] slide 2 [09:00:45] slide 3 - module yang-types [09:01:04] module - inet-types [09:01:48] module ieee-types [09:02:11] example usage [09:03:27] concerned about duplication, have ip-address, InetAddress from smidump conversion, + SMI to XSD [09:05:02] q - maybe data types from ipfix that we could import [09:05:13] dave p - add stuff that you have written into this document [09:05:34] [09:06:24] a - two different things; data types we define for using in yang modules and there is conversion [09:06:41] a - some hacks because we don't have union, which we don't need in NETCONF [09:06:57] dave p - so, there should be duplication, but not where it is semantically different [09:07:25] Issue #1 [09:07:30] Issue #3 [09:07:34] oops #2 [09:07:48] Issue #3 [09:08:32] Issue #4 [09:08:50] Issue #6 [09:08:59] Issue #7 [09:09:34] Issue #8 [09:10:03] [if they type exists in XSD, we should not get inventive] [09:10:07] Issue #9 [09:10:15] Issue #10 [09:11:04] ------------ [09:12:14] outline [09:12:56] What is DSDL [09:14:07] Scope of DSDL for NETCONF [09:17:10] Overview [09:17:20] (sorry, i stepped away) [09:17:28] draft-mahy-canmod-dsdl-02 [09:19:29] draft-lhotka... [09:21:27] q -is code available on web, where martin pointed? [09:21:33] a - created an account at google code [09:21:43] a - people can contribute [09:21:54] dave p - that was the pointer in martin's slidse [09:21:58] googlecode... [09:22:00] a - yes. contributions welcome [09:22:14] "vocabularies used" [09:27:34] bclaise leaves the room: Replaced by new connection [09:27:41] "next steps" [09:29:17] "rpc and notification" [09:33:17] Bal* - not specifically to this. We either had a debate of what the DSDL was describing? The data model or the content. [09:33:30] sharon - all NETMOD thigns need to support both [09:34:41] "Multiple Root Notes" [09:35:59] multiple roots is OK in XSD [09:37:52] Martin - when mapped to NETCONF PDUs, everything will be well-formed. [09:38:30] Martin - goes back to PDU versus content [09:38:48] side issue - standardize container for file storage of NETCONF configurations [09:38:51] Martin - conceptual tree sort of works, but may have other issues [09:38:59] [interesting] [09:39:09] "example conceptual tree" [09:40:01] is this tree needed -- all these YANG definitions are in the same namespace in YANG and XML [09:40:41] Bal* - Need to state even more clearly, how many schema you will generate for every yang module. If I am niave, I would say with one concept. tree, there would be one for everything [09:40:49] Bal* - please me clearer [09:41:17] Lada - You can use it straight with relax ng validators to find errors. DId so with XML examplse from DHCP [09:41:57] [I think it will be interesting to get some of these use cases into Phil's document] [09:42:17] phil - [09:42:31] phil - will it describe the content of edit-config, filters, etc [09:42:40] Lada - right now the same stuff that is expressed in yang [09:43:42] phil - phases of vaidation turned on and off [09:44:14] lada - I can mention that. you can use command line translator to turn on and of features of translation [09:44:47] phil - is there some way to make the content of the RPC section extend a base NETCONF RPC, then add to it with the RPCs defined in new module [09:45:03] lada - that's why I asked for a real-life example [09:45:26] phil - we can do a formal definition. In the DSDL, you descrive the additional tags provided to base model [09:45:51] lada - idea to put information from yang and do input and output. what is done with it later, perhaps out of scope of the DSDL definition [09:46:15] phil - I want to be able to validate [09:46:21] dave p - seee what makes it intot he draft [09:46:31] Q? [09:46:34] -------------- [09:46:51] (back to eye chart agenda) [09:47:37] keep yang and yin in one specification? [09:47:49] no objections [09:48:08] keep XML in the wire in one document for same reason? [09:48:10] no objections [09:48:36] yang datatypes - decided to split out. Any comments? [09:48:43] ready to be accepted as working group draft? [09:48:55] accepted [09:49:30] ray leaves the room: Replaced by new connection [09:50:03] two DSDL drafts - combined into one; once published, not working group draft. [09:50:58] oops. I meant it will be then become working gruop draft, but the next version will be individual submission [09:51:18] ------- [09:51:23] Moving into issues [09:51:57] Juergen's issue list [09:52:09] Issue #1 [09:52:27] Issue #2 [09:52:37] get this feature from N typdefs in a typedef chain [09:53:10] what about multiple error-messages, error-app-tag, etc [09:53:42] already supported in yang [09:54:22] I guess [09:54:34] too complicated IMO [09:56:11] Issue #3 [09:56:12] fine -- N new patterns is easier on the reader than N new typedefs [09:56:57] ray joins the room [09:57:25] B - this may get revved each month. can we handle this? [09:57:38] dave p - this is a risk, but this is a process issue. no good answer [09:58:30] Q - groupings look like type defs. Could ask question. Reusable typedefs for you guys is a good idea .... inheritenace on groupings. Cannot of modify a grouping [09:58:41] dave p - as of today - can't augment a grouping [09:59:14] [10:00:42] dave h - for MIBs, we have a web page that points to typedefs that might be useful to people [10:01:16] Q - need a directorate - MIB doctor type thing [10:02:36] benefit: makes it easier for new modules to import just the reusable bits [10:02:50] Issue #4 [10:04:04] dan - cross participaton, etc [10:04:43] dan - eric grey(gray?) liason to IEEE. Dan liasson on other end [10:06:00] dan - mpls need IEEE, so we enabling IETF work [10:06:03] Issue #5 [10:06:29] Issue #6 [10:07:43] Dave h - comment I missed [10:07:56] [10:08:40] wes - we publish TC in the wrong format. In RFC, that is hard to update, not the way to go. [10:11:53] [10:12:05] a - more an issue when we move beyond the first release [10:12:50] [and I should capture my point about making URLs web accessible] [10:15:35] interim meeting? [10:16:21] Juergen - absolutely needed if we take the deadline seriously [10:16:31] phil - needed [10:16:42] who might attend [10:17:03] count? [10:17:04] 10-12 people [10:17:14] I plan to go [10:17:28] 3 people who can't make it but would like to [10:17:52] what if we move it two weeks later [10:18:26] dan - ask if people who object to the face to face meeting [10:19:04] dan - considering other means, like conferencing [10:19:28] dave p - don't think it's feasible. if author teams, want to use them, that is fine [10:19:50] dave p - time zones make it difficult to be inclusive. [10:20:31] dan - i work 7 time zones away from boss, 10 time zones away from development, it is hard [10:23:55] what about jabber and audio feed from the interim meeting? [10:24:48] actually, it is really bad to mix people in person and those on the audio bridge. [10:24:55] Really. Really. Really. Bad [10:25:19] It's fine here where we have mikes, but when you don't. [10:26:00] what about M-W, 6-8? [10:26:15] what's that? [10:26:24] oct 6-8 [10:26:24] Oh, monday-wednesdsy, the week befroe [10:26:35] we can take it to the list [10:26:45] brabson leaves the room [10:27:22] 4 or 5 prefer virtually [10:27:27] prefer real meeting [10:27:27] twice as many [10:27:47] wes - some of us get to IETF because of other work we do here. [10:28:00] wes - you will get more participation virtually [10:28:41] [again, powerpoint shared works really well] [10:29:03] skype to skype is free [10:29:06] end of session [10:29:14] bye [10:29:22] sharon leaves the room [10:29:34] andy leaves the room: Logged out [10:29:40] DavidAlanReid leaves the room [10:29:44] Juergen Schoenwaelder leaves the room: Disconnected [10:45:10] dlpartain leaves the room [10:47:55] ray leaves the room [11:31:42] Juergen Schoenwaelder joins the room [11:59:59] Juergen Schoenwaelder leaves the room: Replaced by new connection [12:00:32] Juergen Schoenwaelder joins the room [12:57:18] Juergen Schoenwaelder leaves the room: Disconnected [13:47:38] bert leaves the room [14:05:07] bert joins the room [14:15:01] simon leaves the room [14:23:30] lhotka joins the room [14:24:55] Hi Bert, I am just compiling minutes of the session. Not being an experienced jabberist, [14:25:17] can I get somewhere a script from the morning jabber session? Thanks, Lada [15:13:55] bert leaves the room: Computer went to sleep [15:20:46] lhotka leaves the room [16:04:02] bert joins the room [16:08:11] bert leaves the room: Replaced by new connection [16:08:13] bert joins the room [18:41:44] bert leaves the room