[09:56:21] --- bruce has joined
[17:20:33] --- yone has joined
[17:25:05] --- anewton has joined
[17:26:45] --- dblacka has joined
[17:31:49] --- SAH has joined
[17:32:24] --- kjd has joined
[17:32:55] --- markk has joined
[17:33:34] --- hardie has joined
[17:35:49] --- pawal has joined
[17:36:00] --- ggm has joined
[17:36:19] <ggm> ok. so do I scribe or what? isn't that a conflict of interest?
[17:36:29] <ggm> April does Chair "stuff"
[17:36:52] <SAH> nope. knock yourself out.
[17:37:02] <ggm> Milestones.
[17:37:06] <ggm> status uptate.
[17:37:21] <ggm> 3 documents now with RFC-ED [cheers]
[17:37:28] <ggm> thanx to ted.
[17:37:38] <ggm> areg is last big goal
[17:37:56] <ggm> AREG status report, shane and engin
[17:38:41] <ggm> apologies. bunch of changes. did thorough review, engin did his turkish mil. service, so they got held up. talking about stuff you've never seen before
[17:39:25] <ggm> changes to query, results and URI resolution
[17:39:32] <ggm> Query changes.
[17:39:56] <ggm> removed beginsWith/endsWith from contact name searches. goal was to simplify. don't use it today.
[17:40:50] <ggm> separate findOrganizations from findNetworks/findAutonomousSystems. decided searches for resource by name doesnt make sense. current draft doesnt allow searches by net name. can still search on unique ID, handle.
[17:41:14] <ggm> added specificity to findAutonomousSystems. more/less specifics apply here, they are number ranges, have same properties.
[17:41:35] <ggm> removed CIDR (from queries and results) its a special case of start-end, so best handled in the client
[17:41:43] <ggm> Result Changes
[17:41:59] <ggm> no human friendly name or CIDR for nets.
[17:42:22] <ggm> addded URI for network type. noting that "ALLOCATED PA" means what????? <-- [good question -ggm]
[17:42:34] <ggm> removed IP from reverse DNS name server
[17:42:47] <ggm> no human-friendly name for autonomous systems
[17:42:54] <ggm> added parent/noparent for autonomous systems
[17:43:07] <ggm> Ted: when add URI to this, limiting number of schemes?
[17:43:13] <ggm> Ted is mailto: valid URI for this?
[17:43:18] <ggm> Shane thinking HTTP
[17:43:33] <ggm> Ted put it into the spec. people might think its recursive etc. if really for HTTP/HTTPS put into spec
[17:43:37] <ggm> Shane ok., makes sense
[17:44:04] <ggm> Andy wasn't there discussion on list about this? long time ago, first AREG doc enumerated, then we decided not to, now thats not good enough, want URI.
[17:44:49] --- vlevigneron has joined
[17:44:54] <ggm> Shane trying to think of this from user PoV. in general, often, isn't going to want 40 page description of what Allocated PA/PI means, but it doesn't help if they don't have context. specify meaning, specify set of possible types is over specifying, trying to get little more info, Can take to list and discuss. makes more sense?
[17:44:59] <ggm> Andy. string AND URI?
[17:45:06] <ggm> Shane what we're thinking of actually
[17:45:37] <ggm> Nesting
[17:45:43] <ggm> forbid non-nested overlaps
[17:46:14] <ggm> terminology all-less-specifics, one-level-less-specifics, all-more-specifics, one-level-more-specifics, closest-match,
[17:46:27] <ggm> added more examples, with query range. clarify. forgo rest until talked about in WG
[17:46:35] <ggm> Location (URI resolution)
[17:47:05] <ggm> removed text for bottom up resolution. RIRs very nervous. proposing crisp.nro.net (strawman) WKS.
[17:47:33] <ggm> brought notes, will send to ML, about why moving it out of DNS, wiring into protocol. thinking was, unless contentious, wiould go ahead, post to list and get consensus,
[17:47:39] <ggm> Ted seems fragile. why do it this way?
[17:47:49] <ggm> Q? what by fragile?
[17:48:02] <ggm> Ted. no ability to change starting point, short of revolution.
[17:48:15] <ggm> Andy not saying only way, just that its the standard way.
[17:49:09] <ggm> Shane no way to do AS lookups in DNS right now, so have to add namespace to do it.
[17:49:56] <ggm> issue is that there is the octet model, not neccessarily how delegation goes, but namespace follows it. there is a problem with re-delegation, eg with legacy nets split across multiple RIRs, the /16 has gone from IANA -> RIR -> user, delegation has to stop there in DNS.
[17:50:11] <ggm> Ted concern is that it would look like delegation from IANA direct to edge, not able to tell which RIR
[17:50:15] <ggm> Shane for example yes.
[17:51:25] <ggm> also theoretical problem of /7 delegation. have to define protocol to define it as 2x /8, then do less specifics under each. user can enter any IP range, ask for lookup, right now there is no roots, if using DNS it would have to start at in-addr.arpa. does it go to roots? not clear how its going to work. Number of tricky points. going into the DNS doesnt buy much as I see it. some things are nice. cacheing.
[17:51:32] <ggm> Can add that into CRISP, or not.
[17:51:51] <ggm> also political, having org drive this at top.
[17:52:13] <ggm> Markus, dont want to say advantages or disadvantages, can see point with ASNs. can do it with non/8 with SRV and other stuff.
[17:52:45] <ggm> Shane you can, but I dont know from a query user PoV what you gain there. doesnt match real world.
[17:52:54] <ggm> Andy thats an issue. special qualities have to be used.
[17:53:24] <ggm> Shane sorry it didn't go to the list. from RIR perspective, probably the easiest thing to do anyway.
[17:53:31] <ggm> Thats it. not huge set of changes, no major changes.
[17:53:58] <ggm> April. any more? we have some time.. Discussion required on list before you can revamp the draft. once you have a draft we'll do WG last call. how is your time?
[17:54:11] <ggm> Shane what makes more sense? can do that, discuss differences on the list
[17:57:03] <ggm> Ted wasn't speaking ex-cathedra as AD. just personally worried, if this dies, you lose things. it is something we'd want to see in draft as well as mechanism. if going to say 'using this because of <x> eg ASN' it will help.
[17:57:46] <ggm> Shane don't know how much justification text makes sense. I can write 4 on why DNS doesnt make sense. but do we want that for just a one-liner?
[17:57:50] <ggm> Andy well, it doesnt hurt
[17:57:58] <ggm> Shane will make appendix
[17:58:42] <ggm> April. good to have this for IESG review before next IETF. but want to do it before.
[17:59:21] <ggm> Andy. presents on ranges and matches.
[18:00:56] <hardie> Hey, ADs never speak ex cathedra--often ex tempore, or ex animo...
[18:01:01] <ggm> Least, Less, More, Most. Fun.
[18:01:14] <ggm> ex tempore more like it
[18:01:36] <hardie> What, you don't think I'm sincere in my off the cuff remarks/
[18:01:38] <hardie> ?
[18:01:55] <ggm> Andy: the tree view of the network. can be confusing. better to view as address space, showing extents. (horizontal/tabular model)
[18:02:13] <ggm> Andy has marked off the various sub-elements. slides discuss more and less specific
[18:02:43] <ggm> ML discussed lowest level of authority. couldnt find a good definition. if going to use, need a definition. same with highest level of authority.
[18:03:03] <ggm> Q. given two IP addrs, or two networks, if so, then how do you find the two networks given two IP addresses?
[18:03:27] <ggm> thats where some of my confusion comes in. If reading document, had to go implement it, how would you?
[18:03:46] <ggm> Andy hands to Shane
[18:04:21] <ggm> Engin did some ASCII art slides
[18:05:13] <ggm> created a 'virtual' DB of address allocations, with instances of nets sized to fit every case in the required definitional instances.
[18:05:57] <ggm> then can show the queries one by one. start with exact match, entered as start and end address. (which in andys case are the two instances) -and find all exact matches.
[18:06:43] <ggm> all more specifics also is exactly as it sounds, so it finds all the nets 'contained' by the start and end. it also clarifies exact match excludes
[18:08:29] <ggm> one level more specifics limits to the single largest instance (in this case, could be more in others) which is more specific. exlcudes the sub-children. its because of this exclusion that you can define the least and most specific. it wasn't intended to be something a user would see in a query. just used to define this result
[18:08:57] <ggm> all level less specific, does in this case include the 'parent' block.
[18:09:14] <ggm> one level less specific. also limits to the one (first) instance.
[18:09:44] <ggm> these are the core ones. we added 'closest match' -which aims for the exact match, but if there is no exact match, it finds the closest more specific
[18:10:11] <ggm> for routes, finds that route, or the closest matching route. has two examples.
[18:11:12] <ggm> all these examples have been done with unique matches. ARIN does allow non-unique data, exact agreement for start and end. the only way to discriminate, you have to use the handle, to force it to an exact match in lookup terms. all defined in the draft.
[18:11:20] <ggm> Andy. would your DB do parent/child?
[18:11:43] <ggm> Shane no. not allowed. child is the same as more specific.
[18:11:59] <ggm> Cathy. previous versions of the draft, exact match was inclusive in the query.
[18:12:08] <ggm> Engin, yes, true but we changed it to match our current implementation
[18:12:14] <ggm> Cathy breaks ours!
[18:12:25] <ggm> Engin don't know why we have it this way.
[18:12:35] <ggm> Cathy what if they don't provide a start-end.
[18:13:06] <ggm> Shane its start==end. all less specific will see everything in the database, up to 33 nets!
[18:13:22] <ggm> Sanjaya. saying its possible to have these overlaps?
[18:13:32] <ggm> Shane? in APNIC/RIPE no, but in other DBs yes.
[18:16:15] --- hardie has left: Disconnected
[18:19:11] --- sakai has joined
[18:21:44] <ggm> shane discusses corner-case examples with the floor.
[18:23:02] --- norifmi has joined
[18:23:12] --- hardie has joined
[18:23:17] --- kjd has left: Disconnected
[18:26:08] <ggm> query system can't handle some types of query: asking for ranges which have no single match because the span disjoint/nonnesting is probably out of scope
[18:29:00] <hardie> all less specifics with more specifics?
[18:29:37] <hardie> s/more specifics/with more specifics in this range/
[18:29:40] <hardie> ?
[18:31:36] <ggm> Q. sure you can do it by iterations?
[18:31:48] <ggm> Shane. no. not directly.
[18:32:01] <ggm> Q I think multiple searches, multiple round trips, don't know that you got 'enough'
[18:32:27] --- norifmi has left
[18:33:04] <ggm> Andy, mentions IRIS DCHECK
[18:34:54] --- sakai has left
[18:35:30] <ggm> scaled down DREG. no new Q or Res.
[18:35:34] --- hardie has left
[18:35:50] <ggm> not DREG. server ops have flexibility at the s-naptr layer
[18:36:21] <ggm> Q does it do matches or not?
[18:36:58] <ggm> A does 'match' . limits status. can optionally point to real rec. more than just yes or no, but CAN be just yes or no
[18:37:15] <ggm> Q2 who supposed to use?
[18:37:31] <ggm> Andy Use case is fast queries for domain availability. not to replace EPP.
[18:38:07] <ggm> Q2 limited experience, registrants love TCP. don't like UDP, dont want to do UDP query. 500 registrars in .DK, all do TCP whois. not DNScheck.
[18:38:31] --- taiji has joined
[18:38:33] --- taiji has left: Disconnected
[18:38:59] --- taiji has joined
[18:39:07] --- hardie has joined
[18:39:11] --- taiji has left
[18:39:18] <ggm> April. adopted as WG item subject to AD
[18:39:25] <ggm> Shane UDP queries interesting in AREG.
[18:39:31] <hardie> achk
[18:39:31] --- SAH has left
[18:39:32] <ggm> Andy completely reusable.
[18:39:36] --- markk has left
[18:39:42] <hardie> George?
[18:39:46] <hardie> What did you mean?
[18:39:53] --- hardie has left
[18:40:14] --- yone has left
[18:40:29] <ggm> April said "ok, well adopt this work item but its subject to the AD" or at least I *think* she did
[18:40:32] --- anewton has left: Disconnected
[18:40:39] <ggm> s/well/we'll/
[18:41:11] --- dblacka has left
[18:42:07] --- pawal has left: Logged out
[18:47:07] --- vlevigneron has left: Replaced by new connection
[18:47:07] --- vlevigneron has joined
[18:47:08] --- vlevigneron has left
[18:47:21] --- vlevigneron has joined
[18:48:09] --- bruce has left: Disconnected
[18:51:03] --- anewton has joined
[19:12:29] --- anewton has left
[20:05:46] --- vlevigneron has left: Disconnected
[20:11:03] --- ggm has left
[21:29:58] --- vlevigneron has joined
[22:46:51] --- vlevigneron has left: Disconnected
[22:53:51] --- vlevigneron has joined