User:Which Linden/Office Hours/2008 Dec 18

The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

[11:05] Saijanai Kuhn: Donovan went to work for a startup run by a friend. He still posts on the eventlet list though
[11:05] Morgaine Dinova: Oh!
[11:06] Morgaine Dinova: Didn;t know that
[11:06] Morgaine Dinova: Hiya Which!
[11:06] Which Linden: Good morning!
[11:06] Which Linden: How are you all?
[11:06] Saijanai Kuhn: cornfused, otherwise fine
[11:06] Morgaine Dinova: Fine ta. How goes for you?
[11:06] Morgaine Dinova: Oh aye, I'm confused too, hahaha
[11:06] Saijanai Kuhn: mmaybe Bamboo-zled given its your office
[11:07] Which Linden: Heh, what are y'all confused about?
[11:07] Morgaine Dinova: Hehe, you know what. But I guess there's no solution available.
[11:07] Saijanai Kuhn: wel just watchign the interesting reaction from all sides to my suggestion we try to create a universal scripting/plugin API for the SL metaverse clients
[11:08] Which Linden: Oh, yeesh, that is one of the few discussions on sldev that I've actually skipped reading
[11:08] Saijanai Kuhn: sighs
[11:08] Which Linden: Nothing personal, it just seems to be one of those topics, you know
[11:08] Saijanai Kuhn: well, the scripting part is pretty straight forward. Take the Apple vents/AppleScript model of abstracting user actions away from the GUI
[11:08] Morgaine Dinova: Oh yeah, Sai stirred up a hornets nest. Everyone is willing to talk, but only if it's on their turf ;-))))
[11:09] Saijanai Kuhn: all clients handle the "login" command, for example
[11:09] Saijanai Kuhn: the details of implementation are left to the client, but they should all have some response to that comand with a defined set of parameters
[11:10] Which Linden: Yeah, everyone has a different idea of how it must work
[11:10] Morgaine Dinova: Anything prefixed by "Apple" is bad ... unless it's pie.
[11:10] Adz Childs: sees three clouds
[11:10] Which Linden: One upcoming task that I'm going to be helping with is to evaluate message queue software
[11:10] Saijanai Kuhn: forgetting that most of the XEROX team that invented GUIs in the first place eventually ended up at Apple for a few years
[11:11] Morgaine Dinova: That's very interesting Which
[11:11] Which Linden: We're going to be looking at ActiveMQ, RabbitMQ, ejabberd, um, SQS maybe
[11:12] Which Linden: With the goal of having generic queues internally for appropriate use
[11:12] Saijanai Kuhn: ah noted the time. Going into scribe mode. Ctrch you all later as I have to play taxi for bambino
[11:12] Which Linden: See ya sai
[11:12] Morgaine Dinova: See ya Sai
[11:13] Which Linden: Have any of you used message queue systems?
[11:13] Morgaine Dinova: Nope, other than XMPP/Jabberd
[11:14] Aimee Trescothick: no, can't say I have
[11:14] Morgaine Dinova: But they're not generic message queuing solutions are they, any of them
[11:14] Which Linden: The MQ things are more generic than XMPP
[11:14] Lim Catteneo: has some vague memories of borland middlewear from a decade ago or so... mostly nightmarish :P
[11:14] Which Linden: Actually I'm a little unclear what a generic message queue would look like
[11:15] Which Linden: And, yes, this looks like it could be one of those technologies that can be a giant pain in the ass
[11:15] Morgaine Dinova: I'm looking at the AMQP page on WP
[11:15] Lim Catteneo: smells like CORBA ;)
[11:16] Which Linden: Noooooo! COOOORBAAAAA!
[11:16] Which Linden: Actually I don't know much about CORBA (before my time)
[11:16] Lim Catteneo: Which, you are evaluating for possible use in sim<->backend services communication?
[11:17] Which Linden: Lim, yep, purely back-end
[11:17] Which Linden: This is a strong trend at my office hours: discussion of things that Residents will probably never ever see directly or use
[11:17] Aimee Trescothick: lol
[11:17] Which Linden: .... but will be blamed for downtime anyhow
[11:18] Lim Catteneo: well grid stability, if that's the end result of it, is noticable ;)
[11:18] Which Linden: Yes, defninitely
[11:18] Morgaine Dinova: Well the back end is the important en, foundation etc
[11:18] Patnad Babii: so you were saying there is corba in the backend ?
[11:18] Which Linden: Agh, no corba
[11:18] Patnad Babii: ehe sorry my bad
[11:18] Which Linden: :-)
[11:19] Aimee Trescothick: Chicken Corba?
[11:19] Which Linden: Chicken-fried MQ
[11:19] Morgaine Dinova: I'm just wondering whether the market's idea of messaging back end is really the same as you need for SL.
[11:19] Morgaine Dinova: In that you have a pretty unique environment
[11:19] Lim Catteneo: CORBA was the worst programming interface ever, with possible exeception of oracle's OCI ;)
[11:19] Morgaine Dinova: Oh I dunno, I'd say CORBA was streets ahead of UUCP ;-))))
[11:19] Patnad Babii: oh those days ;)
[11:21] Morgaine Dinova: So you're looking at one solution for all needs, or several each appropriate to a task?
[11:21] Which Linden: Well there are a few use cases we're thinking of using this for
[11:21] Which Linden: We definitely don't want this to be a cure-all
[11:22] Which Linden: One useful ability that we could achieve with persistent queues is to abstract away the distinction between online and offline IM delivery
[11:22] Lim Catteneo: do simulator nodes make direct mysql proto connection to the central db now?
[11:22] Which Linden: Lim: yep
[11:22] Which Linden: Which is bad and we'd like to change it
[11:22] Lim Catteneo: poor db ;)
[11:23] Morgaine Dinova: Which, I'm not so sure that that abstraction is a good idea, as the throughput rates are so utterly different.
[11:24] Morgaine Dinova: I should have said latency, not throughput.
[11:24] Morgaine Dinova: You don't care how long email takes to queue. The opposite is true for IM distribution.
[11:25] Which Linden: That's true -- however there are race conditions in the current system that might be avoided if one abstraction handled things
[11:25] Which Linden: Not that person-to-person IM is really especially broken compared to other things
[11:26] Which Linden: Group chat is probably the acid test
[11:26] Saijanai Kuhn: its delivered by the same packet but doesn't seem to lag very often
[11:26] Saijanai Kuhn: and now really MUST run (getting shoes on)
[11:26] Patnad Babii: lag come when there is alot of people into a group mostly]
[11:27] Lim Catteneo: not necessarily
[11:27] Morgaine Dinova: Well, remember that in person-to-person IM, we can't actually *see* the delivery reach destination, so the fact that we've never seen lag doesn't mean much to us --- only to you folks who can see both ends.
[11:27] Which Linden: I believe that person-to-person IM is currently handled in a peer-to-peer way, sim-to-sim, so that's actually kind of a decent architecture, except for the part where it has to decide whether to try and find the person or just stick it in the offline queue
[11:27] Lim Catteneo: yeah, person-person im works fine
[11:28] Lim Catteneo: (most of the time)
[11:28] Morgaine Dinova: Lim: how often do you see both ends?
[11:28] Lim Catteneo: Morgaine, both ends of what?
[11:28] Morgaine Dinova: Of a person-to-person IM
[11:28] Patnad Babii: when chatting to ourself eheh
[11:28] Patnad Babii: with an alt
[11:29] Which Linden: I do that all the time :-)
[11:29] Morgaine Dinova: You chat to your own alt? Well, I guess it takes all sorts ;-)
[11:29] Lim Catteneo: well not that often, but I do follow jira, mailing lists, etc... with a whiny population such as sl, there would be an outcry if it didn't work pretty decent
[11:29] Lim Catteneo: the only problem is when presence is screwed up
[11:29] Lim Catteneo: ie. when sending sim thinks you're offline and sends an IM to your offline queue isted of the live one
[11:30] Patnad Babii: well i can say for a fact that sometime group im are borked not all the time tho
[11:30] Which Linden: Right, yeah, exactly
[11:30] Morgaine Dinova: Lim: there might be no outcry because people IM'ing themselves is so rare :-)
[11:30] Which Linden: Patnad: now there's an optimist!
[11:30] Lim Catteneo: but group chat is what needs to be fixed first imho
[11:30] Lim Catteneo: its the biggest pita in sl experience atm
[11:30] Which Linden: Yeah. The current system has a latency that is directly proportional to the number of members in the group
[11:31] Lim Catteneo: not only that
[11:31] Morgaine Dinova: Yeah. Group IM underpins the social system here. You can't have it not work.
[11:31] Lim Catteneo: it simply does not work some groups
[11:31] Patnad Babii: well i have one hell of a big group 5500 member it work most of the time
[11:31] Lim Catteneo: my understanding is that group servers are distributed on the first few digits of group UUID
[11:31] Which Linden: However, because we don't cap group sizes, we kinda have to evaluate systems that can handle arbitrary numbers of clients for a queue
[11:32] Which Linden: Patnad: online members, I should clarify
[11:32] Patnad Babii: true
[11:32] Lim Catteneo: which means that you can be unlucky and that your group server is way overloaded
[11:32] Which Linden: If most of the members are offline, it's fast :-)
[11:32] Patnad Babii: but this mean almost 100 member always online
[11:32] Patnad Babii: maybe more
[11:32] Lim Catteneo: Which, depends on your UUID first digit luck too ;)
[11:32] Which Linden: Lim: also true; it doesn't load-level
[11:33] Which Linden: That's good to hear Patnad, my experiences with a 250-members-online Linden group has been tolerable as well
[11:33] Patnad Babii: so its a fact of life that cant be changed ? if your group is big cant do nothing about latency ?!
[11:34] Which Linden: Well, we hope to be able to change it
[11:34] Lim Catteneo: group chat system probably needs to be redesigned from scratch ;)
[11:34] Patnad Babii: eheh ok, that would be good also maybe try to design a bigger group capacity then 25 at the same time
[11:35] Which Linden: So, if it turns out that some of these message queuing systems happen to work well for group chat, then it will be parties all around :-)
[11:35] Patnad Babii: lots of people will be happy then
[11:35] Morgaine Dinova: Well live music in SL never recovered since the time that group IM went flakey. It was totally dependent on the Live Music Enthusiasts group IM informing people of gigs, and that totally collapsed. Still dead.
[11:36] Which Linden: :-(
[11:36] Which Linden: Really? They used group IM rather than group notices?
[11:36] Patnad Babii: well if you want to make your group work it can
[11:36] Lim Catteneo: well group notices aren't much more reliable either
[11:36] Lim Catteneo: anything group related is rather flaky ;)
[11:37] Patnad Babii: by designing the message queue system you will also leave space for the OGP to come into play Which ?
[11:37] Morgaine Dinova: Which: notices are used too in LME, but only people with notification rights do it, whereas group IM was used by fans too, repeatedly.
[11:38] Which Linden: Patnad: the message queuing system is completely orthogonal to OGP
[11:38] Which Linden: OGP defines the external interfaces, the details of what services we use internally mostly just affect quality of service
[11:39] Patnad Babii: ok but messages can come from other grid as well right
[11:39] Which Linden: One thing that will be cool about using message queuing software is that we'll have a clear division of responsibilities; one group will be responsible for setting up and maintaining the message queues, while other groups will make applications on top of them
[11:39] Which Linden: Hm, yeah, I have no idea how cross-grid messages happen via OGP -- I think that's not yet defined
[11:40] Lim Catteneo: well nothing much is defined in ogp ;)
[11:40] Patnad Babii: i know but you have to keep this in mind that would be the absolute goal
[11:40] Morgaine Dinova: Not a single LME message has popped up in the last 2.5 hours ... even at this time of day, it used to be full. People just abandoned using it, simply because of the lag and the posting failures.
[11:40] Which Linden: SL tries to be all things to everybody, and ends up half-assing everything
[11:41] Patnad Babii: i had to limit people in my group to 1 message per 10 minutes or most of the people are leaving :D
[11:41] Patnad Babii: so group IM most of the time works
[11:41] Morgaine Dinova: Which: I'd be more generous than that. :-) I think SL was doing just fine ... then it entered non-scalability firefighting, and all was lost.
[11:42] Morgaine Dinova: Ie. 4 years ago we didn't have all these problems. There were bugs, sure, but only silly ones like ghosted objects.
[11:43] Which Linden: You and I will continue to disagree about this, Morgaine :-)
[11:43] Morgaine Dinova: Woohoo! A message just came through on LME ;-))))
[11:44] Lim Catteneo: we're all focusing on problems, the truth is SL has gotten much more stable in the last six months
[11:44] Lim Catteneo: both viewer and the grid
[11:44] Which Linden: That's a great point, Patnad -- even if we could support arbitrary group sizes with arbitrary message rates from each participant, would we want to? It would be spamalicious
[11:44] Morgaine Dinova: Lim: but you didn't see SL when she was truly stable, back in the day.
[11:45] Patnad Babii: well maybe not group > other grid but Person > other grid person would be interesting to have
[11:45] Which Linden: Thanks, Lim, that is my impression as well. I think there's a bit of selective perception, in that since most things work pretty ok, the things that don't stand out more
[11:45] Lim Catteneo: well everybody and their mother sooner or later encounters scalabilty problems
[11:46] Morgaine Dinova: Which + Lim: stop patting SL on the back, you're both newcomers. :-))))) /me chuckles
[11:46] Morgaine Dinova: Yes, true.
[11:47] Which Linden: Heh, yes, and this is something we can't really come to a conclusion on since it depends so much on our personal subjective experiences
[11:47] Morgaine Dinova: But non-scalability is the worst hell to be in, and you must never let it reach you.
[11:47] Which Linden: runs from non-scalability
[11:47] Lim Catteneo: eh i know the feeling
[11:47] Lim Catteneo: you're a startup, need to get something going fast
[11:48] Lim Catteneo: sure you could decide to make it ber scalable
[11:48] Lim Catteneo: but the money guys disagree ;)
[11:48] Patnad Babii: scalability have to be adopted at the early design of every project i'd think
[11:48] Lim Catteneo: given infinite resources yes
[11:48] Lim Catteneo: but rl worlds does not work that way
[11:48] Which Linden: I think it's hard to build in scalability when you're not even sure what it is you're building
[11:49] Morgaine Dinova: It's hell primarily because you can't fix it (short of Bill Gates type money), as it's one of those weakest link in the chain things: you fix one bottleneck, and nothing improves, because the bottleneck just shifts.
[11:49] Lim Catteneo: truth is, scalability is hard :D
[11:49] Aimee Trescothick: Hindsight is always 20/20
[11:49] Which Linden: Sometimes the dimensions along which you must scale differ dramatically from what you'd originally thought
[11:50] Which Linden: As a great example, the designers of SL wanted to avoid the scalability problems faced by world-on-a-server MMOs like WoW or whatever, by splitting the world into regions
[11:50] Morgaine Dinova: Which: are you hinting at something there? :-)))))
[11:50] Which Linden: Now, of course, any individual region cannot handle very many agents, hah
[11:51] Which Linden: Though I suspect that in reality the regions don't have to scale infinitely, we just need to improve the number of avatars per region by a constant amount
[11:51] Morgaine Dinova: Zero said (paraphrasing) "SL will not scale sims, that's a matter for third party grids." If you're suggesting that maybe the thinking on what to scale might be changing .... ;-)))))
[11:52] Which Linden: Well, what I just said is not inconsistent with what Zero said
[11:52] Morgaine Dinova: Pity
[11:52] Lim Catteneo: i still think there must be some room for improvement in the sim code so the performace does not drop so drastically when you have > 40 agents
[11:52] Which Linden: Heh, also we do reserve the right to change our minds
[11:52] Morgaine Dinova: :-)))))
[11:52] Which Linden: Definitely room for improvement
[11:52] Patnad Babii: i heard CG saying they want to expend sim to 512x512 as soon as they have completed the 64bit integration
[11:53] Which Linden: Ha ha, what? I think that was probably a joke
[11:53] Lim Catteneo: its not the size, its the number of agents a sim can handle
[11:53] Patnad Babii: no truly he said that
[11:54] Patnad Babii: you guys should talk more together :D
[11:54] Which Linden: I wasn't there, so maybe he meant it seriously
[11:54] Morgaine Dinova: Well OGP will help scale (but not scalability), since it'll be moving some of the sim's work into the Agent Domain, so the sim will have a bit more CPU available. But it'll still be mathematically non-scalable, while a titchy bit better
[11:54] Patnad Babii: maybe its a dream too
[11:54] Which Linden: It would be a big challenge to implement that, because 256 is hardcoded in many places, not to mention woven into the very grid coordinate system itself
[11:54] Patnad Babii: aparently on opensim you could scale the sim as big as you want tho
[11:55] Aimee Trescothick: 512 x 512 half metres? ;)
[11:55] Morgaine Dinova: Well that's another issue that LL needs to fix: decouple land acreage from pricing.
[11:55] Which Linden: Aimee: and the currency would be the ha'lindendollar
[11:55] Patnad Babii: and having 512x512 region would be interesting for large venue
[11:55] Aimee Trescothick: :D
[11:55] Aimee Trescothick: just scale everything by 0.5 :D
[11:55] Aimee Trescothick: problem solved
[11:55] Which Linden: People, 256 m^2 should be enough for anyone
[11:56] Patnad Babii: sure as simple as that :D
[11:56] Lim Catteneo: lol the ultimate hack
[11:56] Morgaine Dinova: Hehe
[11:56] Lim Catteneo: well i'd much rather see sim being able to handle more prims/agents
[11:56] Lim Catteneo: that to become bigger in area
[11:56] Patnad Babii: right 256 m^2 is big enought to old alot of avatar + a scene and a rock band
[11:56] Morgaine Dinova: But you won't see that. It would need a sim redesign
[11:57] Aimee Trescothick: Yeah, it seems sad to see beautiful huge auditoriums being built ... but you can only ever use a handful of the seats
[11:57] Lim Catteneo: you don'r need to redesign protocol to make sim more effiecient :)
[11:57] Morgaine Dinova: We've had at least a couple alternative designs in AWG for scalable sims, but no interest from LL in them.
[11:57] Which Linden: I seem to recall some chatter that on problem with multiple agents is the ding-dang UDP system and its retries
[11:58] Which Linden: Takes a lot of CPU and bandwidth
[11:58] Morgaine Dinova: Lim: true, but nobody has highlighted inefficiency as a huge problem.
[11:58] Patnad Babii: thats true, current sim can hold a hundred of people but they need to be all cut out of their attachment to give a decent experience
[11:59] Morgaine Dinova: Patnad: OGP will help with that, because in due course, attachments will be handled by the Agent Domain, with the sim only handling bounding boxes.
[11:59] Which Linden: Ha ha, people are even talking about making the protocol more inefficient, with crypto!
[11:59] Morgaine Dinova: LOL, aye
[11:59] Lim Catteneo: yeah
[11:59] Morgaine Dinova: I saw that in SLdev, lol

[12:03] Patnad Babii: or maybe i dont understand what he is doing right
[12:03] Which Linden: Ha ha, sorry I disappeared
[12:03] Morgaine Dinova: It's true, we don't know what M is doing. But the fact that it's shrouded in secrecy is itself a problem.
[12:03] Patnad Babii: welcome back oh lord of the veggies
[12:03] Which Linden: My laptop ran out of batteries because the power connector wasn't in properly
[12:04] Morgaine Dinova: Wb Which :-)
[12:04] Aimee Trescothick: it's what happens when people mention M
[12:04] Lim Catteneo: well M Linden is a businesman, and he seems to show his lack of understanding his customer base every step of the way :)
[12:04] Morgaine Dinova: Hahaha. Your power connector got M'd, Which ;-)
[12:05] Which Linden: Heh, you guys are not fans of M?
[12:05] Lim Catteneo: there is a rare consesus here
[12:05] Lim Catteneo: M is a bad thing for the lab and SL ;)
[12:05] Morgaine Dinova: I liked Philip at the helm :-)
[12:05] Lim Catteneo: too
[12:06] Patnad Babii: i think OGP is really really important for the future of the MV, if we want to become the next internet it have the be with better interop then now
[12:06] Morgaine Dinova: From appearances, M is driving LL back into being a normal company ... ie. closed and retentive.
[12:06] Which Linden: Lack of transparency, ok. Other issues?
[12:07] Lim Catteneo: in other worlds, it appears that he prefers the way of CompuServe
[12:07] Which Linden: (note that I'm not agreeing or disagreeing here)
[12:07] Lim Catteneo: Which, we understand :)
[12:07] Patnad Babii: we wont tell M don't worry :D
[12:07] Lim Catteneo: we see efferots in open source/open standards department getting lower and lower priority at the lab
[12:07] Lim Catteneo: communication with residents is a disaster
[12:08] Which Linden: Are you referring to openspaces there?
[12:08] Morgaine Dinova: Oh, I'm sure M knows how we feel. If he doesn't, then LL *really* has problems.
[12:08] Lim Catteneo: lets not talk about the opesim scandal
[12:08] Lim Catteneo: yeah openspaces is a perfect example
[12:08] Lim Catteneo: many people are offended by "compromise solution"
[12:09] Aimee Trescothick: the cuts in office hours are not a good sign either
[12:09] Lim Catteneo: which means you get even worst product than the original
[12:09] Lim Catteneo: and trying to sell that as an improvement
[12:09] Aimee Trescothick: keep wondering which will be be next
[12:09] Which Linden: hm, really, the current solution with homesteads is perceived to be worse?
[12:09] Morgaine Dinova: Yeah, the withdrawal from the community sent a disastrous message.
[12:09] Patnad Babii: ABSOLUTLY
[12:09] Lim Catteneo: original price hike was just that, price hike
[12:09] Lim Catteneo: homeastead is price hike + sim limits
[12:09] Patnad Babii: more expensive, more restriction where you see better here
[12:09] Which Linden: You realize we were losing money on openspaces, right?
[12:10] Lim Catteneo: i don't belive that
[12:10] Patnad Babii: mkay but why didnt you saw it when it was first offered
[12:10] Lim Catteneo: 16 x 75 x 16 per server
[12:10] Adz Childs: OMG!! you're a potted plant.
[12:10] Patnad Babii: and you decided to change the product after 13 000 unit sold
[12:11] Patnad Babii: generating about 2 and a half millions true USD
[12:11] Lim Catteneo: if you lose money with monthly revenue of 1200 per server, there is something seriously wrong with your company
[12:11] Patnad Babii: thats loosing money ?
[12:11] Patnad Babii: no we are loosing money after all that
[12:11] Which Linden: Yeah, definitely we screwed that up, but we did have to get the prices correct at some point.
[12:11] Patnad Babii: and now you win big time
[12:11] Lim Catteneo: i still don't believe that you lose money at 1200 usd/server/month
[12:12] Which Linden: I haven't run the numbers, but I trust the people who do
[12:12] Lim Catteneo: well you have to
[12:12] Lim Catteneo: we don't .D
[12:12] Patnad Babii: they loose money because 13 000 full sim give more money then 13 000 openspace
[12:12] Which Linden: Yeah, for example I'd probably do the numbers wrong. :-)
[12:12] Morgaine Dinova: Which: there would never have been OpenSpaces if Philip had started work on sim scalability 3.5 years ago when we discussed it with him. OpenSpaces is merely an artifact from the dreadful design of sims --- you had a zillion machines idling because of the terrible architecture, and you tried to replace the needed technological solution by marketting. It failed, surprise surprise.
[12:12] Lim Catteneo: if by lose you mean "make less that we could" sure
[12:13] Which Linden: I understand your perspective, Morgaine, definitely respect that
[12:13] Morgaine Dinova: Well so many things arise from that problem, it's just sad for me to see it.
[12:13] Which Linden: And I don't necessarily disagree that a more fluid system should be our engineering goal
[12:13] Which Linden: It's actually something I'd love to tackle myself
[12:14] Lim Catteneo: now with 125/homestead still running 16/server gives you whopping 2000 usd / server which is outrageous ;)
[12:15] Which Linden: Anyhow, I gotta run, need to get in gear on fixing some bugs. Thanks for stopping by!
[12:15] Adz Childs: B-bye!!! :D
[12:15] Lim Catteneo: thanks for listening to all these rants lol
[12:15] Morgaine Dinova: Thanks for the chat Which :-)
[12:15] Aimee Trescothick: :)
[12:15] Which Linden: I appreciate your honesty :-)
[12:15] Patnad Babii: okay which sorry we kinda derailled form the subject :D
[12:15] Which Linden: And your eloquence :-)
[12:16] Morgaine Dinova: Yeah, we ranted a lot, sorry Which. It's frustrating :-)))
[12:16] Which Linden: BTw, next week I won't make it to my office hours.
[12:16] Morgaine Dinova: kk
[12:16] Which Linden: In fact, my next office hours will be January 8!
[12:16] Which Linden: Have a happy new year!
[12:16] Aimee Trescothick: ooh, yes, crikey, it's that time of year
[12:16] Lim Catteneo: see you next year Which, happy holidays!
[12:16] Aimee Trescothick: happy christmas and new year!
[12:16] dogtow Hand: thx happy new year we all hope :o)

User:Which Linden/Office Hours/2008 Dec 18

Navigation menu

Search