[08:50:29] ilari.liusvaara joins the room [08:54:33] Carsten Strotmann joins the room [09:02:46] audio is good [09:04:18] Olafur Gudmundsson joins the room [09:04:41] Homnet starting (your friendly scribe) [09:04:53] Mark talking about agenda [09:05:26] Brian Carpenter joins the room [09:05:30] Markus Stenberg joins the room [09:05:46] equinox joins the room [09:05:57] equinox is now known as David Lamparter [09:05:58] Ted Lemon joins the room [09:06:36] Tim Chown starting architechture document presentation [09:06:36] Andrew Sullivan joins the room [09:06:44] William Lupton joins the room [09:06:46] 70 discusses iin IESG [09:06:58] to large extent done [09:07:27] section 3.5.1 rewritten (see presentation) [09:08:02] Dave T. with out saying name, something about mime types [09:08:24] does anyone have a link to the slides? [09:08:36] O&M 3.8.2 was short, has been expanded [09:08:53] the section was unclear still undergoing reservations [09:09:31] wants to start new draft on O&M needs volunteer, talk to chairs if interested [09:09:36] Slide 5 [09:09:51] New routing protocol work goes to routing area [09:10:02] (read slide for details) [09:10:11] slide 7 [09:10:35] bortzmeyer joins the room [09:10:52] On multiple routing protocols (words by Adrian Farrel) [09:11:31] Lee Howard: do we need require single or is the requirement to be able to route to each other [09:12:11] Mark: said lots of stuff Lee: What you said is not in the text [09:13:01] Lee: will send text to list [09:13:30] Dave Thaler: agrees with first point disagrees with second point (on slide 7) [09:14:02] MarkT: needs to be explicit and agrees with DT [09:14:23] Laurenco: spectaular copout, need to pick one and only one [09:14:54] Mark closes line [09:15:38] MarkT: as long as 0 is an option DT and LH are ok [09:15:56] slide 9: complete view [09:16:19] Lee Howard joins the room [09:16:38] MIchalel ??: (mumble) [09:17:07] Dave Thaler: this is requiing that the routing protocol is a link state protocol [09:17:27] Fred Baker: this constrains the soluiton space why ? [09:18:06] Slide 10: LLN's [09:18:13] slide 11: conf info [09:18:55] slide 12: last slide default deny 3.6.1 (controversial???) [09:19:22] Laurenco: COntent free [09:19:24] text [09:19:39] (Lorenzo Colitti) [09:20:13] Slide 13: Publishing will push out version RSN (today or tomorrow) [09:20:20] hopefully IESG will approve [09:20:54] Daniel Migault: Homenet Naming DHC Options [09:21:20] Homent Front End Naming Arch. [09:21:25] (slide 2) [09:21:58] outsource the public DNS serivicing so CPE is not answering queries [09:22:18] CPE is the master, public servers are secondaries [09:22:40] slide #3: [09:23:23] cpe (Master and signer) — zone transfer —> public servers [09:23:31] Slide 4: [09:23:35] slide 5: [09:23:36] cpe doesn't sign it, at least slide indicates it does dns2dnssec on ISP side [09:25:01] Mark Andrews: does not trust ISP to have keys [09:25:54] Thomas W: do you trust cpe to do keying [09:26:40] Ian Farrer joins the room [09:26:40] Mark A: answers question how the CPE does it [09:27:16] Ray: SHould ISP sign, or user/CPE sign? Mark: ISP should not sign my data; they are not authoritative. [09:27:26] Ray: clarification question but isp is using their key [09:28:34] Ted: 2 usage cases, one where the CPE is trusted and they can do it but if CPE is not trusted this model is better [09:29:02] Stewart: Agrees with Mark [09:29:43] all that data is coming home is subject to tampering thus signing at source is better [09:29:55] Ray: WHat is sig(0) is used [09:31:18] Stewart: if the home gateway is compromised all bets are off, having the CPE sign makes it only one one point of attack [09:31:31] Ole Tango Romeo Oscar Alpha November (TROAN) [09:31:44] ????: how do you handle multihoming [09:32:27] ???? = Ole Troan [09:32:27] Will this work when one does not work (Actually Ole T) [09:32:47] with the ISP but using 3 party [09:33:26] Mark A: will work if ISP is willing to serve the zone but it is up to user use or not use [09:33:45] Andrew S: is confused, [09:34:13] William Lupton leaves the room [09:34:42] Andrew Sullivan: What is not clear how are you handle the split-DNS when there are two possible key that sign the same data, public and private views [09:34:50] William Lupton joins the room [09:36:04] ???: has problem with that CPE are throwaway devices [09:36:13] Ralf Weber [09:37:55] Michael Richardson: CPE signing is much better in particlar case when an ISP is served a warrant to make changes [09:38:13] Slide 5: DHCP option [09:38:39] The bottom line in all this is that we _really_ need to get some kind of consensus on naming—we can't just go with Daniel's draft, but we don't have anything else except my dns-pd draft, which doesn't solve the whole problem. [09:38:43] slide 7: [09:39:18] slide 8: Name server set options [09:39:45] personally I see this as somewhat too complex solution for one case [09:39:51] slide 10: Scopes [09:40:10] Lee Howard leaves the room [09:40:15] but I guess it depends on what is the goal, really [09:40:52] Part of what I would like to see as an outcome is that people actually have naming in the home, and for people who aren't geeks like us that requires substantial support from the ISP. [09:41:07] I think a significant part of the goal here is to create more services managed directly by the ISP [09:41:08] This is a solution that appeals to ISPs. But we also need to allow for a solution that's more in line with what Michael is talking about. [09:41:17] slide 12: zoro conf [09:41:22] zorro! [09:41:42] zoro ==> zero [09:41:49] Ted, We _have_ those solutions. When you set up your AP at home, you get a drop down list with my employer and a bunch of our competitors in it, and you register there. [09:41:52] slide 14 [09:42:03] Slide size 15: [09:42:22] I am _not_ arguing that we should just leave that to the market, but it seems to me that standardizing such a 3d party arrangement mechanism would be somewhat better [09:42:26] You mean dyndns? That's incomplete—no PTR records. [09:43:05] FWIW, I don't think that Dyn's system or anyone else's is adequate. PTRs are the least of the problems — huge amounts of the reverse space hasn't been maintained. [09:43:20] Mark: Send oppiions to the list real important topic need to nail this down soon [09:43:21] There are lots of other problems — how to do DNSSEC in such a model is just as bad [09:43:23] Ole Troan joins the room [09:43:39] Juan-Pedro Cerezo Martin joins the room [09:43:44] Yes. [09:44:09] Anyway, we keep talking past each other, and we seriously need to converge on this. [09:44:17] But I think the model that is not linked so tightly to DHCP would be better. But I agree about that. [09:44:25] Home network control portoc; Steven Barth [09:44:32] slide 2: [09:44:38] DHCP is just a mechanism for getting the bits where they need to be. Shouldn't be the only mechanism. [09:44:47] But it is convenient, and will be available in many cases. [09:44:53] Sure [09:45:04] Lee Howard joins the room [09:45:10] Tried to implement OSPFv3 to gain experience [09:45:30] Ted, I agree it is convenient, but then you'd need a separate protocol between your internal DNS server and the CPE [09:45:48] There's a big assumption encoded in that statement, Ole. [09:45:54] and what if someone you don't have a "DHCP Path" to acts as a secondary, couldn't we use the same protocol [09:45:59] But yes, if your internal DNS is not running on the CPE, you are correct. [09:46:21] I just want this solved for my home, and that's how it is running. ;-) [09:46:32] Lee Howard leaves the room [09:46:33] DHCP works to communicate configuration information from the ISP to the CPE. I think internally HNCP may be a good choice—not sure yet. [09:46:58] For homenets we need a DNS solution that can happen automatically relying solely on home routers, without setting up an additional server. [09:47:07] Mark Townsley joins the room [09:47:16] Lee Howard joins the room [09:47:30] Agree, that also needs to be supported. But remember external naming requires "configuration" of some sort regardless. [09:47:48] Andrew Sullivan leaves the room [09:47:51] ???: you can shorten the time to converge [09:47:54] You mean at least a whitelist so that all the machines in the home aren't visible to the world? [09:47:58] That's Acee Lindem. [09:48:07] sorry, can somebody tell me where on the agenda we are? [09:48:20] We're on HNCP. [09:48:32] David Lamporter. [09:48:36] thanks [09:48:42] p [09:48:43] np [09:48:43] ???: ospf not that difficult ? [09:48:51] Olafur, that was David. [09:48:58] He's apparently a quagga hacker. [09:49:35] 1) Not all homes will have an external zone. 2) The name has to be "created" and there has to be a bilateral agreement between EU and ISP about that name. 3) Only services that wish to be externally visible will register in the global namespace. [09:49:41] sdie: Data sync [09:49:53] Yup. [09:49:53] s / sdie/ slide [09:51:11] Slide: Data model [09:51:34] (please put slide numbers on slides for scribes) [09:53:21] Andrew Sullivan joins the room [09:53:49] Slide: Security [09:54:49] Slide: Payloads, Extensions, Glue [09:56:02] Ole: I agree with your 3 points [09:56:53] form my point of view, Ole's thing mandates mostly manual step SOMEWHERE anyway [09:56:54] Slide: Discovering Borders [09:57:00] so I'm not sure why DHCP comes in here in the first place [09:57:14] I wouldn't want my first-hop ISP to do more than just delegate the reverses to DNS provider of my choosing [09:57:24] (possibly more than one ISP, obviously) [09:57:41] How do you communicate that to your ISP, Markus? A web UI? That's not automatic. [09:57:46] then again, even for clueless home user case, I'm not sure having whole zone visible is bad [09:57:58] Ted Lemon: probably giving them money for premium service [09:58:15] Markus: For "cluless user" why have anything visible ? [09:58:17] considering the 'default' service set of ISPs is kind of .. sigh. [09:58:17] It seems prudent to make the option to control what is published available. [09:58:50] well, depends how optimistic are about end device security, but if network security is sane you can just not show whole clueless zone [09:58:55] as it's probably behind firewall anyway [09:59:34] Slide: Routing Protocols [09:59:36] the architecture doc says (in three paragraphs) that we don't have a position on whether firewalls are enabled or disabled. [10:00:05] optimistic people would turn off firewalls and advertise full zones in clueless case [10:00:20] Sure. [10:00:27] Notetaker requests everyone have slide numbers on their slides in the future. [10:00:28] Slide: Routing selection [10:00:54] But really for the most part I would expect that the end user will request specific publications, not be aware of the idea of general publication. [10:01:35] That is, they might want to access some device on the homenet when they are away, but will not think to open up the firewall entirely, nor to open up the name publication entirely, because they aren't strongly aware of these concepts. [10:01:38] yeah, I'd imagine that router having UI to prod naming service provider with stuff, and then notify with something would be useful [10:01:42] I'm not sure if DHCP is the way to do it, though [10:01:56] It's not _the_ way, but it is certainly _a_ way. [10:02:06] agreed [10:02:25] Just because you can do something doesn't mean you should, however [10:02:54] the history of "[DNS|DHCP] is there, so we should use that" is not a glorious one [10:02:57] Barbara Stark: you talked about having algorithm for determining external do you have one for internal ? [10:03:24] VoIP over TXT record, anyone? [10:04:31] sometimes the "Wan" port can be internal and how to check that [10:04:32] Ole Troan leaves the room [10:04:53] Juliusz Chroboczek [10:05:05] Juliusz: implement is good, draft is not [10:05:15] needs update and clarification [10:05:19] Ole Troan joins the room [10:05:48] Juliusz: is the draft overspecifying ? too many MUST's [10:06:39] Not everything is supposed to be mandadory [10:07:33] Ralph Schmieder joins the room [10:07:43] Juliusz: is this a general purpose configuration protocols for all networks not just homenets ? lie does it allow /128 configuration [10:08:36] [10:09:24] Juliusz: the document assumes the confiuration deamon is assumed to be root (wrong) [10:09:41] root —> routing [10:10:10] answer: do not want multiple routing protocols someone has to select [10:10:10] Andrew Sullivan leaves the room [10:10:22] Andrew Sullivan joins the room [10:11:20] Acee Lindem thought the doc made him infer a lot, wants more detail [10:11:23] Acee: not draft overspecfying, needs more text on trickle had to use his experience to configure [10:11:58] Erik: what sort of testing did you do with ISP's ? [10:12:27] Not much testing with ISP's needs real world testing [10:13:07] askign for input of userclass option [10:13:52] Acee: on DHCPv6 are you also redistributiing the source doing that otherwise you will never converge [10:14:28] not really [10:14:38] Mark: similar to MIF problems [10:15:16] ????: DHCPv6 in priniple all router have this info [10:16:12] Laurenco: mixed interfaces makes this complicated [10:16:40] Mixed interfaces should be treated as external [10:17:26] Markus Stenberg leaves the room [10:17:29] John (comcast): do not refuse packets by default users try strange things [10:18:00] Marcus: DNS-SD/MDNS talk [10:18:18] Slide 2: Service discovery [10:19:07] Slide 3: What is needed [10:19:49] Andrew Sullivan leaves the room [10:20:25] Andrew Sullivan joins the room [10:20:58] Slide 4: [10:21:14] How can it be zeroconf ? [10:22:32] Slide 5: question [10:22:34] Question: In this solution, has Apple TV AirPlay mirroring been tested using DNS-SD with unicast DNS through the hybrid proxy? [10:22:50] is that for the mic? [10:23:03] Kerry Lynn: great work [10:23:03] yes, please. [10:23:26] will do [10:23:43] is this protocol agnosit >>> answer yes [10:23:59] routing protocol agnostic [10:24:09] thanks, Andrew. [10:24:20] lol [10:24:29] Andrew for Ralph S: apple TV, Marcus does not have applie TV no but buy me one [10:25:53] Stewart: Apple TV is a special case (TV, Music compicate apple TV), legal reasons make Apple TV have restrictions on sharing that he personally agree with. [10:26:14] Tim: asks for feedback on DNSDS draft [10:26:40] Markus Stenberg joins the room [10:26:44] Markus Stenberg leaves the room [10:26:47] Markus Stenberg joins the room [10:27:02] Pierre Pfister: Prefix and address assiginments [10:27:22] sllide 2: Motivations [10:27:55] Andrew Sullivan joins the room [10:27:59] sadly enough I think the earlier comment also applies to itunes, think it works only on same link [10:27:59] Andrew Sullivan leaves the room [10:28:08] (probably due to same contractual reasons, I'm sure) [10:28:46] Slide 2: Introducing the "Flooding protocol" [10:29:54] Slide 3: Adding IPv4 Support [10:30:08] Correction Slide 4: [10:32:25] Slide 5: Other Additions [10:32:56] Olafur Gudmundsson leaves the room [10:33:21] Olafur Gudmundsson joins the room [10:33:41] Slide: Some more details [10:35:09] Slide 7: What doesn't change [10:35:26] Slide 8: Thank [10:35:47] www.homewrt.org [10:37:03] Laurenco: two comments on the new additions, old draft was stateless , this statefull, you need additional complexity to sync DHCP databases [10:37:28] Laurenco: either do DHCP-failover or remove [10:37:49] How to do DHCPv4 —> NAT [10:38:48] Marcus: this will work and not cause problems, in most cases [10:39:05] Laurenco: there are times when this will not work in all cases [10:39:25] John Brzozowski, didn't identify himself at the mic [10:39:31] John B: bad idea to do more NAT [10:39:36] Ole Troan joins the room [10:39:57] Ole Troan leaves the room [10:40:21] Ole Troan leaves the room [10:40:27] MarkT: we were chartered for v6, we are trying to define v4 congruent approach [10:40:52] Laurenco we are not considering the failure modes enough and that means we can not recover [10:41:27] Ole Troan joins the room [10:42:17] Laurenco: Does not want have DHcpv5 with state without failover [10:43:15] MarkT: take to list [10:44:03] Ted: (as techical contibutor) we can use behave techology to satisfy Laurenco [10:44:16] Acee: agrees with Laurenco [10:44:26] Lorenzo, by the way. [10:44:41] I am also opposed to DHCPv5 state. :) [10:44:49] Sorry for all the misspelled names :-( [10:45:15] Ted: what RFC defines DHCPv5 ? [10:45:21] No criticism intended. [10:45:39] DHCPv5 is documented in RFC3315i. [10:45:55] (where i is the square root of negative 1, for those who aren't aware) [10:46:37] Timothy WInters: Homenet design team [10:46:55] http://tools.ietf.org/agenda/89/slides/slides-89-homenet-6.pdf [10:47:01] Sldie 2: [10:47:07] SLide 3: Updates to draft [10:47:47] Slide 4: SPER to Homenet [10:49:23] Slide 5: Homenet to 7084 [10:50:19] Slide 6: Multi-layer SPER [10:50:23] (dones not work) [10:51:01] Can you detect it ? answer NO [10:51:53] Mark T: Choosing a solution path [10:52:53] . 1.  Go back to working on OSPF for rouKng and configuraKon . 2.  Adopt HNCP for configuraKon and minimalist rouKng now, “full-blown” rouKng TBD . 3.  Propose something else [10:53:18] Ole Troan leaves the room [10:53:31] we need to move to solutions as archtechture is done [10:54:05] wants feedback are we ready to go work on solution space and figure out what routing to use [10:54:39] Open Mike [10:55:12] Juliusz: Clarified point 1 [10:55:17] Ted Lemon leaves the room [10:55:30] Ole Troan joins the room [10:56:26] Lee howard: has problem with OPSF it is painfull and comlexty that he is not sure we can subject on the world — he is leaning towards something HNCP [10:56:38] Lee: SHould pick one RP [10:57:57] Michael A: wants see techical analysis on the various protocol on handling large TLV's before picking solution [10:58:29] MarkT: not sure we want to do that [10:58:39] will take forever [11:00:09] Brian Carpender: Not sure the arch doc is helpful documnent to select some anlysis is needed [11:00:26] Lee Howard leaves the room [11:00:42] BC: not number #3 can we do it with #1 as new takes longer [11:01:49] Barbara Stark: just one solution, keep it simple !!!!!!!! PICK ONE (repeated few times) [11:02:36] BS: #2 will lead to TBD never happen [11:03:06] Dave T: supprorts #2 needs running code before specificatin [11:03:24] Ole Troan leaves the room [11:03:53] Michael R: echo DaveT, need to understand what did not work with OSPF [11:04:31] Argument against HNCP : it has no Wikipedia page [11:05:13] MR: HNCP may has to decide if routing on or off, all the picking devices get turned on and off and that may affect behavior of #2 [11:05:41] Lee Howard joins the room [11:06:12] Marcus: hard to have homenet running w/o routing protocol [11:07:02] ????: with OSPF we did not discuss how to receive DHCP options [11:07:52] ????: HNCP is doing the same as OSPF, i.e. just reinventing the wheel, OSPF problems are perceptions [11:08:43] JohnB: #3 is off the table, likes #2 as it allows incremental deployment [11:08:56] JohnB: agrees with Barbara to keep it simple [11:10:31] Acee: Agrees with Marcus, OSPF is used in many environent and lots of issues with adding things to it, —> HNCP might be better [11:11:09] Lorenzo: Routing protocol must be used and the right number is 1 [11:11:44] L: HNCP we can move faster than getting the OSPF gods to bless any changes [11:12:51] L: HNCP can do both routing and configuration and that is attractive, worries about fragmention on OSPF, leaning towards #2 [11:13:15] Ole Troan, "Tango Romeo Oscar Alpha November" [11:13:54] Ted: no opinion [11:14:41] Brian??: keep it solution OSPF is too complex, wants RIP [11:15:32] Ted Lemon joins the room [11:16:04] Juliusz: we know how to debug a configuration protocol, others know how to debug routing protocl, noone knows how to debug combined protocol [11:16:53] J: likes RIP as routing protocol, OSPF is a little more effort —> wants single routing protocol [11:17:21] Ole Troan joins the room [11:17:30] ?????: Mixing routing and configuration leads to crazy code base [11:17:38] Henning Rogge [11:17:51] Henning: likes #2 [11:18:21] Acee: with HNCP you can define scopes, you can not do that with OSPF or any other routing protocol [11:19:36] Mark: wants sense of the room #2 is not exactly what to [11:20:26] Dave T: hard for people to determine if variants of #2 are #3 [11:21:21] Ted: ask the questions but make it clear that #2 is not final [11:21:49] Andrew Sullivan leaves the room [11:22:09] Dave T: ask first about conf then routing [11:22:30] Lee: how to vote for TBD [11:22:59] Mark: hum for OSPF for configurtion [11:23:05] hum for HNCP [11:23:10] hum for something else [11:24:11] MarkT: strong supoort for HNCP weak for TBF [11:24:34] MarkT: Hum for OSPF only [11:25:02] Curtis: OSPF first and something else [11:25:17] restart hum [11:25:31] Ted: is there a hum for I do not know yet [11:26:38] MarkT: can we addopt HNCP as wg draft ? [11:27:17] Ray: we can have 2 options for hum; Do we want to support 0 or 1 [11:27:31] Hum for support 2 or more [11:27:43] Ray: More support for 0 or 1 [11:29:22] Michel A:wants delay in deciding on adoption of HNCP [11:29:45] Ted: not ready to hum on HNCP Routing thing [11:30:43] MarkT: we need a working group draft on HNCP, is this the approach ? [11:30:49] hum now [11:30:49] Ian Farrer leaves the room [11:31:15] not deiced [11:31:21] not adopt [11:31:37] Ted Lemon leaves the room [11:31:38] Mark stong for support for addopt, some for do not know, none against [11:31:43] Ole Troan leaves the room [11:31:44] Ole Troan joins the room [11:31:44] Ole Troan leaves the room [11:31:45] End of meeting [11:31:49] Olafur Gudmundsson leaves the room [11:31:50] Mark Townsley leaves the room [11:31:55] Lee Howard leaves the room [11:32:01] Markus Stenberg leaves the room [11:32:09] Carsten Strotmann leaves the room [11:33:18] Ralph Schmieder leaves the room [11:33:32] Juan-Pedro Cerezo Martin leaves the room [11:35:37] Brian Carpenter joins the room [11:35:49] Brian Carpenter leaves the room [11:36:09] bortzmeyer leaves the room [11:36:19] ilari.liusvaara leaves the room: offline [11:52:13] David Lamparter leaves the room [11:53:20] bortzmeyer joins the room [12:05:55] Lee Howard joins the room [12:13:01] William Lupton leaves the room [12:13:32] Lee Howard joins the room [12:13:57] Lee Howard leaves the room [12:38:57] Ted Lemon joins the room [12:40:18] Ted Lemon leaves the room [12:43:08] Ted Lemon joins the room [12:47:55] Brian Carpenter leaves the room: offline [12:48:57] Lee Howard leaves the room [12:49:28] Lee Howard joins the room [12:50:16] Lee Howard leaves the room [12:53:27] Ted Lemon leaves the room [12:59:40] Ted Lemon joins the room [13:00:57] bortzmeyer leaves the room [13:01:55] Andrew Sullivan joins the room [13:02:01] Andrew Sullivan leaves the room [13:02:06] Lee Howard joins the room [13:02:38] Ted Lemon leaves the room [13:03:14] Lee Howard leaves the room [13:03:20] Markus Stenberg joins the room [13:03:33] Markus Stenberg leaves the room [13:06:40] Ian Farrer joins the room [13:08:17] Mark Townsley joins the room [13:09:56] Mark Townsley leaves the room [13:11:48] Mark Townsley joins the room [13:13:21] Ian Farrer leaves the room [13:13:31] Ian Farrer joins the room [13:21:25] Mark Townsley leaves the room [13:23:53] Mark Townsley joins the room [13:25:42] Mark Townsley leaves the room [13:36:06] Olafur Gudmundsson joins the room [13:45:21] Ian Farrer leaves the room [13:47:43] Ian Farrer joins the room [13:58:49] bortzmeyer joins the room [14:01:48] bortzmeyer leaves the room [14:02:15] bortzmeyer joins the room [14:06:22] Olafur Gudmundsson leaves the room [14:12:39] bortzmeyer leaves the room: Replaced by new connection [14:12:39] bortzmeyer joins the room [14:20:35] Olafur Gudmundsson joins the room [14:33:22] Ian Farrer leaves the room [14:37:53] Ole Troan joins the room [14:38:19] Ole Troan leaves the room [14:43:56] Ian Farrer joins the room [14:56:59] Olafur Gudmundsson leaves the room [15:26:10] bortzmeyer leaves the room [15:26:50] bortzmeyer joins the room [15:29:52] Ole Troan joins the room [15:30:57] Ole Troan leaves the room [15:33:23] Ian Farrer leaves the room [15:57:05] Ole Troan joins the room [15:59:06] bortzmeyer leaves the room [15:59:47] Ole Troan leaves the room [16:10:30] Ian Farrer joins the room [16:11:58] bortzmeyer joins the room [16:43:48] Ian Farrer leaves the room [17:49:56] bortzmeyer leaves the room [17:55:13] bortzmeyer joins the room [18:12:43] bortzmeyer leaves the room [23:04:00] bortzmeyer joins the room [23:51:57] bortzmeyer leaves the room