Simulator User Group/Transcripts/2012.06.29
|Prev 2012.06.26||Next 2012.07.03|
List of Speakers
|Andrew Linden||Ardy Lay||Baker Linden|
|Davido Chrome||Kallista Destiny||Kelly Linden|
|Long Xu||Michi Lumin||Moundsa Mayo|
|Rex Cronon||Sahkolihaa Contepomi||Simon Linden|
|Susan Tsuki||TankMaster Finesmith||Valinye|
[16:03] Rex Cronon: hi moundsa
[16:03] Susan Tsuki: Simon, I added an agenda item on https://wiki.secondlife.com/wiki/Simulator_User_Group
[16:03] Moundsa Mayo: And hello Andrew!
[16:03] Andrew Linden: Hello.
[16:03] Rex Cronon: hi andrew
[16:03] Susan Tsuki: Hi, Andrew, how are you?
[16:03] Baker Linden: Hi everyone
[16:04] Rex Cronon: hi baker
[16:04] Andrew Linden: Hrm... is that agenda up to date Simon?
[16:04] Simon Linden: I haven't touched it
[16:04] Andrew Linden: I've never actually pulled items from the wiki agenda.
[16:04] Davido Chrome: It still have my item from about a year ago...
[16:05] Andrew Linden: actually, I didn't know someone had added the agenda section 8-0
[16:05] Simon Linden: I think some of the agenda is out dated, but let's get to the new things today and then maybe clear it out
[16:05] Susan Tsuki: then you get your pick of priority!
[16:05] Andrew Linden: Tell you what Susan, why don't you just state your agenda item here, in a couple minutes once we open the table for submissions.
[16:06] Susan Tsuki: Sure!
[16:06] Susan Tsuki: Proposal: Return the user stipend with a two-bracket progressive income tax, where anyone making x% above median income pays y% on the excess, which is distributed in equal proportion to everyone with less than L$z who has logged in over the past w days, where (x, y, z, w) are optimized for economic growth.
[16:06] Susan Tsuki: the title is "Why is there no sim called Krugman?" but that's just a title
[16:07] Andrew Linden: Oy. You want to add taxes back into the system?
[16:07] Davido Chrome: I think you just gave me a headache.
[16:07] Kallista Destiny: Income?
[16:07] Susan Tsuki: only on the highest incomes
[16:07] Rex Cronon: lol. tax?
[16:07] Andrew Linden: We had a tax/stipend system a long time ago.... but it was based on the amount of objects you had in-world and your avatar's reputation.
[16:07] Davido Chrome: Yeah, I would hit Minus pretty fast...
[16:07] Kallista Destiny: Income? What bloody income, I spend money here
[16:08] Susan Tsuki: more important is to return the stipend to save the company and, with France, set an example of good governance to the world
[16:08] Andrew Linden: Ok Susan, we'll revisit your proposal in a bit.
[16:08] Susan Tsuki: thank you
[16:08] Andrew Linden: In the meantime... news, if we have any.
[16:08] Simon Linden: I can talk about the releases ...
[16:08] Andrew Linden: Go ahead Simon.
[16:09] Simon Linden: The last of the OS updates rolled on Wednesday, so we're all on the new system
[16:09] Kallista Destiny starts humming the International and waving a red flag
[16:09] Davido Chrome: Oh, that's why it's so glitchy, new system.
[16:09] Simon Linden: .. also using some new deploy tools. It will be interesting how that goes in the next few weeks - hopefully the rolls get quicker
[16:09] Andrew Linden: New operationg system: debian/squeeze
[16:10] Simon Linden: The crash rates we've seen under the new OS are good
[16:11] Simon Linden: We did have an interesting problem with the rolls as we lost an unexpectedly high number of machines as they were powered off and back on -- normally that doesn't happen, so some of these boxes hadn't been turned off in a long time
[16:12] Simon Linden: It's not uncommon to have hardware failures, since the grid is so large, but this was bigger than expeted
[16:12] Simon Linden: expected too
[16:12] Sahkolihaa Contepomi: Ouch.
[16:12] Rex Cronon: next time don't turn them off and on?
[16:12] Davido Chrome: Maybe this will means some hardware upgrades?
[16:12] Andrew Linden: We probably had uptimes around two years or more for some of those machines.
[16:12] Simon Linden: We _think_ it's a one-time problem - the machines that were about to die did, and the ones that made it are pretty solid
[16:13] Andrew Linden: Actually, the OS upgrade was a prerequisite for getting new hardware.
[16:13] Simon Linden: We are working on getting new hardware - the next machines are pretty amazing
[16:13] Kallista Destiny: This is a know problem , machines are most likley to die on power down/up
[16:13] Sahkolihaa Contepomi: Mostly hard disk drive failures, or full motherboards?
[16:13] Davido Chrome: Oooh, nice!
[16:13] Davido Chrome: Because of power spike?
[16:14] Ardy Lay: Could be various little things keeping them down.
[16:14] Simon Linden: I'm not sure, Koli - I haven't seen that level of detail. I'm not sure they're checking too deeply - these were mostly old machines that we'd replace in the upgrades anyway
[16:14] Andrew Linden: Simon, will there be rolling updates next week?
[16:14] Simon Linden: Next week there will be a main grid roll on Tuesday, and that's it
[16:15] Davido Chrome: Will the new machines be 64 bit?
[16:15] Davido Chrome: Is SL turning 64 bit?
[16:15] Simon Linden: Yes, Alve
[16:15] Kallista Destiny: Every roll will involce an reimaging correct?
[16:15] Sahkolihaa Contepomi: I doubt the sim needs more than 4GB RAM.
[16:16] Simon Linden: I think we're looking at dual hex-core systems - 12 cores in each one :)
[16:16] Davido Chrome: Does this mean 64 bit viewers?
[16:16] Simon Linden: No, it's unrelated to viewers
[16:16] Andrew Linden: (4GB of RAM should be enough for anybody...)
[16:16] Sahkolihaa Contepomi: Say that to us multi-taskers. ;)
[16:16] Kelly Linden: We are not, at this time, moving simulators to 64 bit.
[16:16] Kallista Destiny: Viewers are independant of what the server is running
[16:16] Simon Linden: ah, I mis-understood your 64 bit question
[16:16] Kallista Destiny: mostly
[16:17] Davido Chrome: I have 4 gigs. Only time I ran out of memory was when I tried to render a lwan of grass flowing naturally in the wind in Blender...
[16:17] Sahkolihaa Contepomi: I ... have 16GiB. 4.8GiB used.
[16:17] Ardy Lay: I don't think I have quite needed a 64-bit viewer yet but that's partially because I set the LAA flag on the 32-bit viewer.
[16:17] Davido Chrome: Kelly, my fault. I asked two similar questions in succession. XD
[16:18] Andrew Linden: The new servers will have much more than 4GB, I believe.
[16:18] Simon Linden: much, much more, and a nice SSD
[16:18] Davido Chrome: It would be nice on Linux. It might mean less hassle getting voice and in world youtube to work.
[16:19] Simon Linden: Converting to 64 bit is a painful project because of all the libraries we work with
[16:19] Davido Chrome: 64 bit viewer I mean.
[16:19] TankMaster Finesmith: ugg, sorry im late
[16:19] Kallista Destiny: 64bit viewer seems to be a bit of a pain too, and there really sin't a lot of gain, and some loss....
[16:19] Davido Chrome: Simon, it will have to happen some time though?
[16:20] TankMaster Finesmith: what loss?
[16:20] Kallista Destiny: fetching 64bit operands takes longer that 32 bit operands
[16:20] Simon Linden: They use up more memory due to the pointer size
[16:20] TankMaster Finesmith: no they dont
[16:20] TankMaster Finesmith: (refering to taking longer)
[16:20] Ardy Lay: Pointers are big. ;-)
[16:20] Sahkolihaa Contepomi: I wouldn't expect to see the simulators using 64-bit until one region manages to go over 3GiB and crash.
[16:20] Andrew Linden: We made an attempt to go to 64 bit servers several years ago
[16:20] Davido Chrome: Not talking aboput mouse pointers I take it?
[16:21] Andrew Linden: but at that time we were running into some simulator memory footprint problems
[16:21] Ardy Lay: 64-bit mouse pointers would be large too.
[16:21] Susan Tsuki: it's not a good idea to compare by space or time alone. Use memory bus bandwidth instead.
[16:21] Andrew Linden: and going to 64bit pushed us even further on the footprint
[16:21] Andrew Linden: so we punted that transition into the future
[16:22] Andrew Linden: but some benchmark experiments suggested that we might actually gain a 10-20% speedup going to 64bit
[16:22] Susan Tsuki: it's like trying to get double floats inside quaternions: too many things need to change for it to be an easy transition?
[16:22] TankMaster Finesmith: not suprising, you have access to twice the number of registers
[16:23] Davido Chrome: Would be less susceptible to lag from too many scripts in region, perhaps?
[16:23] Susan Tsuki: similarly, if you go to 64 bit hardware, you only get the gains you've planned for and everything else can slow down
[16:23] Andrew Linden: I think the pain of going to 64 bit today would mostly be around rebuilding all the libraries we need
[16:23] Davido Chrome: (GUessing here...)
[16:23] Kelly Linden: Unfortunately moving to 64bit for the simulator would effect script memory sizes for mono scripts.
[16:24] Andrew Linden: but I expect we'll try it again, especially if we think there would be some performance improvements
[16:24] Susan Tsuki: s/effect/affect/
[16:24] Kelly Linden: So it will need to be done carefully. :)
[16:24] Andrew Linden: ah, I didn't know that kelly
[16:24] TankMaster Finesmith: that would be fixable though, right kelly?
[16:24] Sahkolihaa Contepomi: Heh, mass crash from stack-heaps.
[16:25] Kelly Linden: Andrew: We would need to run them as 64bit as well which would double the pointer size. In many cases the change would be insignificant but in some pedantic cases it could cause memory exhaustion.
[16:25] Andrew Linden: Ok, any more news?
[16:25] Kallista Destiny: pathalogical programs perform poorly
[16:26] Andrew Linden: Ok, so the table is open.
[16:26] Andrew Linden: Susan has a proposal to bring back tax and stipend.
[16:26] Andrew Linden: Susan, were you around when we had the tax/stipend/reputation system (2003)?
[16:26] Davido Chrome: For High income people!
[16:26] Susan Tsuki: No, but bond rates in France are down 0.2% since Sarcozy lost to a constituion-amending Socialist supermajority. The free market has spoken.
[16:26] Kallista Destiny: I move that Susan be drawn and quartered.
[16:27] Susan Tsuki: I only got here in 2006
[16:27] Rex Cronon: this isn't france? right.
[16:27] Andrew Linden: Hrm. Well the tax/stipend system wasn't very popular.
[16:27] Kallista Destiny: Nyet
[16:27] Susan Tsuki: human behavior is human behavior
[16:27] Andrew Linden: It was a game mechanic that was imposed on the world.
[16:27] Kallista Destiny: and is not Soviet Union either
[16:28] Susan Tsuki: of course it wasn't popular! Of the sims near Da Boom named after economists, they are almost all named after Austrian school economists
[16:28] Susan Tsuki: but it's a terrible mistake
[16:28] Rex Cronon: actually we r taxed already. each time u change $L to real money ll takes a cut:)
[16:28] Susan Tsuki: it's causing the stock price to slowly collapse
[16:28] Susan Tsuki: and concurrency to decrease
[16:29] Susan Tsuki: remember when LL was proud to publish economic statistics of Linden dollar transactions?
[16:29] Davido Chrome: I live in a socialist country. It works really well...
[16:29] Susan Tsuki: slowly they were retracted, as governance changed to obfustication and deciet
[16:29] TankMaster Finesmith: i dont really see what the advantage whould be
[16:29] Andrew Linden: Susan, when you say "stock price"... what are you talking about?
[16:30] Meeter: Timecheck : User Group is half over
[16:30] Susan Tsuki: Rex has a point: the tax today is flat, which is regressive. Why aren't you stimulating growth instead?
[16:30] Susan Tsuki: Andrew: the market pricing of expected revenue
[16:31] Rex Cronon: just to make it sure. i don't support a tax system.
[16:31] Davido Chrome: I might be mistaken, but aren't the Lindens here Tech guys? Not economists having anything to do with SLs economy?
[16:31] Kallista Destiny: There is no tax in Second life, except that imposed on the people in the EI (VAT)
[16:31] TankMaster Finesmith: theres a tax on items sold on the marketplace
[16:32] Susan Tsuki: Kallista: that is also flat and regressive
[16:32] TankMaster Finesmith: for items costing over a certian amount
[16:32] Moundsa Mayo: Well, Rex, if certain of the values of x,y,z, and w were == 0 I'd support the tax concept.
[16:32] Simon Linden: Very true, Alve - we're interested in how things run, but we're very far from the Linden economic management
[16:32] Long Xu: any date on when getregionagentlist will work on huds for parcel detection?
[16:32] Ardy Lay: I move that economic issues are out of scope for this venue.
[16:33] Kallista Destiny: the Sl exonomy id dpressed, becasue the world exonomy is depressed, People have less disposable income to fritter waya on games.
[16:33] Rex Cronon: this tax thing should be brought up at one of those economy group meeting. oops, there r no such meetings
[16:33] Simon Linden: Errico - do you know of a jira for that?
[16:33] Long Xu: i am sure kelly knows of the problem
[16:33] Rex Cronon: right. make a jira:)
[16:33] Kelly Linden: Errico: It is in the next server maintenance project. (simon: it is in DRTSIM-145)
[16:33] Long Xu: woo
[16:33] Susan Tsuki: Linden Lab stock isn't even listed on https://welcome.sharespost.com/companies-on-sharespost
[16:33] Susan Tsuki: anymore
[16:33] Davido Chrome: Any news on when Pathfinding rolls out?
[16:34] Rex Cronon: gets votes for it, and have WorkingOnIt Linden work on it;)
[16:34] Andrew Linden: Davido, we attempted to put pathfinding up on RC this week, but it had to be rolled back.
[16:34] Susan Tsuki: oh I'm sorry, it's still there as a logo, they just don't show the price anymore
[16:34] Davido Chrome nods, Ok. What happened?
[16:34] Andrew Linden: We've fixed that bug, and a few more, but next week is a no-change-window on the RC channels.
[16:35] Andrew Linden: So we'll try pathfinding again the week after that, if there is an open slot.
[16:35] Susan Tsuki: http://taterunino.net/statcharts/mo_cmedian400full.jpg shows the stagnation problem
[16:35] TankMaster Finesmith: TPVs are still several weeks away from starting to adopt mesh
[16:35] Davido Chrome: Mesh?
[16:35] Michi Lumin: o_O
[16:35] Andrew Linden: The bug that blocked pathfinding was that you couldn't move regular objects across region boundaries at all.
[16:35] Ardy Lay: Heh..
[16:36] Sahkolihaa Contepomi: I assume you mean pathfinding, Tank. :p
[16:36] Valinye: I was wondering what happened... Region came up, 'wheee pathfiding!', server went down again, 'awwww no pathfinding.'
[16:36] Michi Lumin: was gonna say, i was pretty sure that t-shirt had already been purchased.
[16:36] TankMaster Finesmith: er PF
[16:36] Davido Chrome: Aaah, that explains why my cross region teleports didn't work for a while.'
[16:36] TankMaster Finesmith: not mesh :P
[16:36] Ardy Lay: Golly gee, Radegast even renders mesh objects and many still think of it as a text-only client. ;-)
[16:37] TankMaster Finesmith: ugg, long day
[16:37] Ardy Lay: Hehe
[16:37] TankMaster Finesmith hates going into viewer release mode
[16:37] Davido Chrome: Tankmaster, wuill FS wait with their release untl PF comes out?
[16:37] Andrew Linden: Yeah, there is some work to be done for TPV's to be ready for pathfinding.
[16:37] Sahkolihaa Contepomi looks at his clock and sees 12:38AM' "Pft."
[16:37] TankMaster Finesmith: yes, davido, and we dont have havok yet
[16:37] Andrew Linden: We also get a chance to fix some of the minor known bugs that didn't quite make it to the first RC.
[16:38] Kallista Destiny: Umm andrwe if I understood Oscar it was between PF anf non PF and vice versa. due to the different fromat of the binary encoding of the object in the two different Havock versions.
[16:38] Andrew Linden: Kallista, you mean the fact that objects couldn't cross region boundaries?
[16:39] Davido Chrome: Makes sense. I live in a Sl server region, adn the adjacent one where my store is, and my TPs go to, is RC.
[16:39] Kallista Destiny: Yes Andrew
[16:39] Andrew Linden: Yes, there will be an asymmetry in how _mesh_ objects can cross region boundaries once PF goes to RC
[16:39] Andrew Linden: you'll be able to move from non-PF to PF with mesh objects, but not the other way
[16:40] Kallista Destiny: Oscar was muttering about have contenet wids RC channels for mesh changes, just because of that.
[16:40] Davido Chrome: Ah, But prims will work for some reason?
[16:40] Andrew Linden: however the blocking bug was that non-mesh objects were also being blocked from crossing region boundaries in that direction
[16:40] Andrew Linden: and they should have been unaffected
[16:40] Davido Chrome: Aaah.
[16:40] Andrew Linden: there was a bug in the logic (and it was actually code that I had changed, so it was my fault)
[16:41] Ardy Lay offer Shrek Ears to Andrew.
[16:41] TankMaster Finesmith: heh, know how you feel, andrew
[16:41] Kallista Destiny: However I was able to tosse some daggers from non PF to PF and vice versa, however the dggers were temp-on-rez.
[16:41] Davido Chrome: But this will be a nonissue, because pathfinding will go grid wide ones it has been tested on Release Candidate servers?
[16:41] Kallista Destiny: Once the two regions could see each other again.
[16:42] Kallista Destiny curses the region presence bug.
[16:42] Andrew Linden: Right Davido. Once everything has the same version of Havok (the libs for the physics engine) all meshy region crossings will work.
[16:42] Ardy Lay: Oh wait, the Shrek Ears are just for build blockers.
[16:43] Davido Chrome whispers: Ah, yes. Has there been any changes since you fixed the script collision report problem
[16:43] Davido Chrome: ?
[16:43] Davido Chrome: Ah, yes. Has there been any changes since you fixed the script collision report problem?
[16:43] Andrew Linden: it used to be that when an LL developer would break something so that many others would have to spend time to fix it, they would wear some Shrek Ears to display their shame
[16:44] Simon Linden: Alve - not sure what you mean ... changes in what?
[16:45] Andrew Linden: Davido, was that a script collision report problem in pathfinding? or the main channel?
[16:45] Davido Chrome: https://jira.secondlife.com/browse/PATHBUG-56
[16:46] Davido Chrome: You guys fixed it. would like to know there are no regressions.
[16:46] Andrew Linden: That was in May. We changed everything since the ;-)
[16:46] Susan Tsuki: Kallista: the graphs on http://dwellonit.taterunino.net/sl-statistical-charts-testing/ show that the LL economy is decoupled from the world economy. Compare http://taterunino.net/statcharts/cbands0.jpg to http://research.stlouisfed.org/fred2/series/EMRATIO -- the cusp is years apart and at a different angle
[16:47] Davido Chrome: Oh. Better make a simball arena in a pF sandbox and test again then...
[16:48] Andrew Linden: Actually, theose two graphs do correlate around 2008
[16:49] Andrew Linden: Of course, the SL graph is going to have a different timescale. That graph starts in 2007 or so.
[16:49] Ardy Lay: Gambling and advertising are included in the world economy.
[16:50] Michi Lumin just isn't sure what this has to do with the railroad. c_c
[16:50] Davido Chrome: Isn't gambling against the TOS?
[16:50] Susan Tsuki: the http://taterunino.net/statcharts/cbands0.jpg peak is in January 2009, sorry,
[16:50] Andrew Linden: yeah, 2009. I was going on memory and forgot to examine the timestamps on the bottom carefully.j
[16:50] Ardy Lay: I kinda thing pathfinding railroad might be interesting.
[16:51] Michi Lumin: indeed, ardy.
[16:51] Ardy Lay: -thing +think
[16:51] Davido Chrome: Would it allow for smoother turns?
[16:51] Kallista Destiny: The Caledon Railroad will benefit greatly
[16:51] Andrew Linden: Kitto was trying to get multiple disconnected cars to cross region boundaries in sequence, but we still have some work to do on region crossings.
[16:51] Susan Tsuki: but the only cash flow graph at the end of http://dwellonit.taterunino.net/sl-statistical-charts-testing/ which could be said to have peaked did so closer to January 2010
[16:51] Moundsa Mayo: Have experimented with PF rolling stock.
[16:52] Moundsa Mayo: The Aditi SLRR Test Regions may be placed on PF channel. Fixed track ther to test and develop with.
[16:52] Davido Chrome: How much script time and lag increase will the pathfinding stuff add, if any?
[16:53] Ardy Lay: Maybe trolleys would make more sense in a world where flexible yet tensile couplings cannot be constructed reliably.
[16:53] Andrew Linden: Davido, very little. The "character" feature set is very performant.
[16:53] Davido Chrome: Nice.
[16:53] Valinye: Pft, trains. Matter transport is the wave of the future!
[16:54] Andrew Linden: You can bog the region down with characters, but the time is spent in the physics engine rather than scripts.
[16:54] Andrew Linden: We had 10's and 100's of characters running early on with suprisingly good performance.
[16:54] Davido Chrome: Ah, what about the grid mesh, will that affect performance anything?
[16:54] Andrew Linden: I haven't looked lately.
[16:55] Andrew Linden: The pathfinding feature only works once the "navigation mesh" (navmesh) has been computed.
[16:55] Meeter: Timecheck : User Group is almost over
[16:55] Andrew Linden: And it takes a lot of CPU cycles to compute the navmesh
[16:55] Andrew Linden: but it is done in a separate thread, in the background
[16:55] Davido Chrome: But once done, it's there?
[16:55] Andrew Linden: it can take 10's of seconds, maybe even a couple minutes to compute
[16:55] Andrew Linden: depending on how complicated the region content is
[16:55] TankMaster Finesmith: multi-threading is nice
[16:56] Andrew Linden: yeah, once it is computed the characters can find their way with relatively few CPU cycles
[16:56] Davido Chrome: Will people driving physical objects make the nav mesh update continuosly?
[16:56] Sahkolihaa Contepomi: Shame havok doesn't have 'native' GPU support like Physx does.
[16:56] Andrew Linden: Yes, moving objects have to "cut" the navmesh in real-time so that characters can avoid such obstacles.
[16:57] Andrew Linden: So the more dynamic objects means more time spent in the physics engine
[16:57] Davido Chrome: Is this true even if no pathfinding charachters are avtive?
[16:58] Davido Chrome: active?
[16:58] Andrew Linden: Yes
[16:59] Andrew Linden: We may try to allow the navmesh stuff to be disabled on a per-region basis.
[16:59] Davido Chrome: Then the last question is, I think, can region owners turn off the Nav mesh?
[16:59] Davido Chrome: Aah.
[16:59] Susan Tsuki: I've updated https://wiki.secondlife.com/wiki/Simulator_User_Group#Agenda with buttressing links. I'll make a Jira out of it and take my chances.
[17:00] Meeter: Thank you for coming to the Server User Group
[17:00] Andrew Linden: Actually, I just asked Falcon. He says that option is already done
[17:00] Andrew Linden: and will be accessible to Estate Managers via the simconsole window.
[17:00] TankMaster Finesmith: its in the estate consol
[17:01] Andrew Linden: yup
[17:01] TankMaster Finesmith: having a clickable option in preferences seems like a more discoverable way
[17:01] Davido Chrome: Aah, nice. I know Simball regions will want to turn that off, very physics intense game.
[17:02] TankMaster Finesmith: just a side note, mesh adoption is now up to almost 92% of the loging
[17:02] TankMaster Finesmith: logins*
[17:02] Andrew Linden: Yes TankMaster, it should eventually be put in the World --> Region/Estate UI
[17:02] TankMaster Finesmith: i agree with that
[17:02] Kelly Linden: Have a good weekend all!
[17:02] Davido Chrome: You too!
[17:02] TankMaster Finesmith: you to kelly
[17:02] Andrew Linden: Ok. Thanks for coming everyone.
[17:02] Rex Cronon: tc kelly
[17:03] Rex Cronon: tc andrew
[17:03] TankMaster Finesmith: tc simon, andrew
[17:03] Moundsa Mayo: Andrew, Simon, Kelly, Baker, thanks for your patience, your time, and all your hard work!
[17:03] Simon Linden: Thanks everyone for the good chat -- see you next time
[17:03] Sahkolihaa Contepomi: Thanks, Lindens.
[17:03] Davido Chrome: Thank you very much for your patient answering of questions!
[17:03] Rex Cronon: tc simon
[17:03] Ardy Lay: Thanks for having us
[17:03] Sahkolihaa Contepomi heads off to bed.
[17:03] Rex Cronon: tc everybody, and have a nice day
[17:04] Michi Lumin: Thanks.. for sure
|Prev 2012.06.26||Next 2012.07.03|