[09:09] Mic Bowman: g'morning justin [09:09] Justin Clark-Casey: Hi Mic [09:09] Aaron Duffy is Offline [09:09] Doug Osborn: aloha justin :) [09:10] Justin Clark-Casey: hi doug [09:10] Mic Bowman: justin... can you take a look at this and tell me what you think... [09:10] Mic Bowman: http://www.sciencesim.com/wiki/doku.php/foundationnotes [09:10] Doug Osborn: and mic you look familiar :) [09:10] Mic Bowman: so long as i don't look gray [09:10] Robert Adams is Offline [09:10] Robert Adams is Offline [09:10] Aaron Duffy is Online [09:10] Doug Osborn: i just descided to go with it and went to the grey hair :) [09:11] Mic Bowman: well... grey avatar hair is fine [09:11] Aaron Duffy is Offline [09:11] Robert Adams is Online [09:11] Mic Bowman: 50+ avatars on wright plaza yesterday? [09:11] Doug Osborn: that was the news :) [09:12] Mic Bowman: all the changes that have been applied seem to be working really well [09:12] Aaron Duffy is Online [09:12] Doug Osborn: and its good news too [09:12] Mic Bowman: chamomile is down under 1.5G of memory [09:12] Doug Osborn: :) [09:12] Lightswitch: lights on [09:13] Mic Bowman: aaron isn't here... but we have an idea for a patch that would really help him with his memory problems as well [09:13] Aaron Duffy is Offline [09:13] Mic Bowman: the technique we used to optimize the ODE mesh management can be used for script assemblies as well [09:13] Justin Clark-Casey: mic: very interesting. It looks like there is still some debate as to the aim. You thin kthat viz/edu is a broad enough field? [09:13] Justin Clark-Casey: yes, the 50+ was very impressive considering the instability of the past 3 months [09:13] Mic Bowman: justin... for sciencesim i think its the right place to start [09:14] Aaron Duffy is Online [09:14] Mic Bowman: if the community becomes sufficiently active it will expand [09:14] Justin Clark-Casey: I see you're talking both distribution and grid - you think one sufficiently synergizes the other to be worth the effort? [09:15] Mic Bowman: i'm not sure... [09:15] Doug Osborn: i think its the right direction Mic [09:15] Mic Bowman: when we get to the point that CB is "complete" then the differences are largely irrelevant [09:15] Doug Osborn: stability is somethjing I know justin likes :0 [09:16] Mic Bowman: stability is good if the focus is on apps sitting about the platform [09:16] Aaron Duffy: Hello all [09:16] Mic Bowman: a wise man (my phd advisor) commented to me once that you can only innovate at one layer at a time [09:16] Justin Clark-Casey: yes - I want to see novel apps and the more stability the more such activity will occur [09:17] Doug Osborn: lol [09:17] Mic Bowman: the "platform" has to become boring for real innovation to take place in the applications [09:17] Aaron Duffy: Robert, I saw your interview on Visualize Intel [09:17] Mic Bowman: Robert... our new star!!! [09:17] Robert Adams: blush [09:17] Justin Clark-Casey: I see no funding issues are mentioned yet :) [09:18] Mic Bowman: we've been having a number of discussions about funding [09:18] Mic Bowman: but we were talking funding without having a clear idea of what we were funding [09:18] Mic Bowman: that ordering is somewhat.... complex [09:18] Justin Clark-Casey: ha :) [09:18] Mic Bowman: i figured we should go back and clarify what we want the foundation to actually do [09:19] Mic Bowman: aaron... [09:19] Mic Bowman: if you could finish swimming... [09:19] Aaron Duffy: you can see me? I've been a cloud all morning [09:19] Mic Bowman: what is the current scalability limit on your fern growth? [09:19] Justin Clark-Casey: the two interlink though. In one way the broader the aims the more opporunity for funding I imagine. But it's probably a bell curve - get too broad and it starts to look less achievable and too diffuse [09:19] Mic Bowman: justin: yes [09:20] Justin Clark-Casey: hence I was interested if the cost of maintaining a grid is paid back by the attraction of extra interest :) [09:20] Mic Bowman: we've talked funding... time to talk purpose... and then back to funding [09:20] Aaron Duffy: over about 2500-3000 scripted objects in the region at once and it gets too laggy to do anything. [09:20] Justin Clark-Casey: uh huh [09:20] Mic Bowman: laggy for the viewer or for the simulator [09:21] Mic Bowman: justin: i'm still working on what "grid" means [09:21] Aaron Duffy: I haven't tested with latest improvements but I was having thread-related crashes after about 16000 objects have been created and deleted [09:21] Mic Bowman: thread or memory? [09:21] Justin Clark-Casey: philosophically or in terms of terminology? [09:21] Aaron Duffy: laggy as in, the region fps drops to nothing [09:21] Mic Bowman: hmm... [09:21] Doug Osborn: mono scripts? [09:21] Aaron Duffy: running out of threads, regardless of my MONO_THREADS_PER_CPU or xengine settings [09:21] Justin Clark-Casey: mic: it would be good to see more detail about the role of a distro - presumably not just to package opensim core but also to provide useful extras on top for viz/edu purposes [09:22] Mic Bowman: justin: some of both... CB lets us move away from osgrid and reactiongrid... [09:22] Robert Adams: there seems to be several ways for OpenSim to run out of threads [09:22] Aaron Duffy: and really I only get to 16000 rezzed and deleted by NOT using llDie. I delete dead objects using DeleteSceneObject from a region module. Wth llDie you can only rez and delete 3400 [09:22] Mic Bowman: in that services like assets/inventory are no longer part of the grid [09:23] Mic Bowman: or at least bound to the grid [09:23] Mic Bowman: and the user server can be made transparent [09:23] Mic Bowman: that really takes you back to the map [09:23] Justin Clark-Casey: mic: Is that getting into experimental territory? Not so much synergy with stability (though potentially very interesting things are possible with a popular distro) [09:23] Mic Bowman: and "trust" [09:23] Mic Bowman: possibly... [09:24] Mic Bowman: but the innovation is still in the way the region simulators are used [09:24] Mic Bowman: the rest is really about making the platform boring [09:25] Mic Bowman: we want to get to the point where someone can deploy a region, connect it to a data feed, and see the results with little more effort than starting an http server and writing some php [09:25] Mic Bowman: and some of the usages... like teaching a class... should be as simple as downloading a pre-configured simulator that loads an oar file [09:26] Justin Clark-Casey: that sounds like a good core aim. In my experience with open source projects it's very important to have a clear and achievable vision. One liners like that one appeal to me [09:26] Mic Bowman: one liners can hide all the "interesting" detail [09:26] Mic Bowman: :-) [09:27] Doug Osborn: justin.... Mice mentioned somethingin a previous meeting that i always wanted to ask about.... does the SVN prim saving stuff actually still work? [09:27] Doug Osborn: Mic* [09:27] Justin Clark-Casey grins [09:27] Justin Clark-Casey: Doug: I would guess not, it hasn't been touched for a very very long time [09:27] Justin Clark-Casey: more than a year at least [09:27] Mic Bowman: is there an easy way to configure out the optional modules? [09:27] Doug Osborn: whats the feedback been on the IAR files? [09:28] Justin Clark-Casey: Beyond setting them as disabled = false on the ini [09:28] Robert Adams: you get rid of modules by deleting the dlls... they are loaded if they are in the bin directory [09:28] Justin Clark-Casey: doug: not much yet. Still some little bugs to iron out, and I had to compromise their usefulness for political reasons, which was annoying [09:29] Justin Clark-Casey: robert: yes [09:29] Doug Osborn: and Mic where does work like Justin IAR fit in the stable spectrum.... we really need that kind of functionality :) [09:29] Mic Bowman: IAR is as stable as anything else... [09:29] Justin Clark-Casey: doug: also, the advantages over using things like SI and Meerkat is less clear cut. SI is simpler (?) but iars do the whole object inspection so you get all the object contents [09:30] Mic Bowman: to be clear... my notion of stable is closer to predictable [09:30] Doug Osborn: Im assuming since he has an "API" we count "assume th" the bendend stuff would ber there [09:30] Justin Clark-Casey: doug: well, you know, what would be really useful is if you try and just file some bugs [09:31] Doug Osborn: yes that is the way :) [09:31] Rob Willis: jcc i saw something about miniregion modules on your blog, how powerful can they be compared to the normal [09:32] Justin Clark-Casey: rob: actually that's not an area I'm very familiar with yet. It's worth reading Adam's blog for more details on that [09:32] Justin Clark-Casey: I haven't seen a lot of public use yet [09:32] Rob Willis: ok [09:32] Mic Bowman: i've done a bit with MRM [09:33] Mic Bowman: it's pretty cool but the API is still pretty thin [09:33] Doug Osborn: and hows the megaregion testing going Mic? [09:34] Mic Bowman: yellowstone & st helens are both running as megaregions [09:34] Mic Bowman: you should only teleport into the SW region [09:34] Mic Bowman: and i crash the region about half the time I teleport out [09:34] Mic Bowman: or crash the viewer [09:34] Doug Osborn: Shen and I heard an interesting pitch yesterdaI ve been in yellowstopne :) [09:34] Doug Osborn: oops [09:35] Doug Osborn: i went to yellowstone without any issues a couple time :) [09:35] Mic Bowman: i don't think we have all the megaregion patches... [09:35] Mic Bowman: but i think we're close [09:35] Mic Bowman: its a VERY cool experience... [09:35] Mic Bowman: now i think we need to put 130K prims in a 3x3 or 4x4 megaregion [09:35] Aaron Duffy: or 10000 scripted objects [09:36] Mic Bowman: with the memory optimizations we should be able to do that without too many problems [09:36] Mic Bowman: yup [09:36] Mic Bowman: don't think we aren't going to try that!!!! [09:36] Mic Bowman: we're trying to get some of the networking code stablized first [09:36] Mic Bowman: speaking of which... [09:36] Doug Osborn: well Im wodering what you can do with the surrounding regions [09:37] Mic Bowman: conversation with jhurliman yesterday... the plan right now is to wait for another two weeks and then pull opensim head into scisim [09:37] Mic Bowman: opensim head has most of our performance patches right now [09:37] Mic Bowman: and it has john's texture and htb code [09:37] Mic Bowman: we just need to get permission to release the scene priorization code [09:38] Justin Clark-Casey: I'm still coming across patches in the Mantis which have no comment of being applied (though they may have been). [09:38] Mic Bowman: the one patch that removes the sleep should probably not be applied right now [09:39] Mic Bowman: there are too many other race conditions causing problem [09:39] Mic Bowman: we removed it and it hasn't caused any significant problems... but the comment on the sleep is a bit disconcerting [09:39] Justin Clark-Casey: http://opensimulator.org/mantis/view.php?id=4080 ? [09:40] Mic Bowman: and region loading is fast enough... yes 4080 [09:40] Justin Clark-Casey: thanks [09:40] Mic Bowman: its sad that we use sleeps to avoid race conditions... [09:40] Mic Bowman: but also necessary at tiems [09:40] Justin Clark-Casey: it's hacky [09:41] Mic Bowman: yup.. [09:41] Mic Bowman: so one other comment that came up in the senate meeting... [09:41] Mic Bowman: as you all know, we have a git repository with current scisim code [09:42] Mic Bowman: it should be possible for us to provide commit access for those who would be willing to contribute to the code [09:42] Mic Bowman: would that be useful? [09:43] Mic Bowman: one approach we could take is to allow for creation of branches and then only work through merges on the head [09:43] Doug Osborn: yes it probably would...although i dont know how long it will bee before I get to read CB code... [09:43] Aaron Duffy: at my level of coding, some method of approval before commit would be wise [09:44] Mic Bowman: i'd rather avoid a significant overhead... but i'd also like to extend the scisim development group outside intel [09:44] Doug Osborn: I dont know enough about GIT management yet to comment on the best approach to manage [09:44] Doug Osborn: so Id be happy to go with waht you like [09:44] Justin Clark-Casey: I don't think I have anything useful to contribute yet [09:44] Mic Bowman: i do have a bit of experience with software dev :-) [09:45] Mic Bowman: i was thinking that the first thing we should do is get a copy of extsim into the distribution [09:45] Mic Bowman: in the addons directory [09:45] Justin Clark-Casey: extsim? [09:45] Mic Bowman: i think i could also put the quicksort region module there [09:45] Mic Bowman: extsim is the 3di region module for connecting backend data sources easily [09:46] Justin Clark-Casey: ahhh [09:46] Mic Bowman: in fact... now that i think about it... the quicksort module would be a great test case [09:46] Mic Bowman: one other thing i would like to see... [09:46] Mic Bowman: is some improvement in the configuration methodology [09:47] Mic Bowman: something that can make set up easier [09:47] Justin Clark-Casey: I didn't know extsim was even active, no files were released on gforge [09:47] Aaron Duffy: The three .ini system is a bit confusing [09:47] Mic Bowman: we have a bunch of scripts we use to manage our regions that others might find useful [09:48] Doug Osborn: with a distribution you can make a number of "decisions" to support your relaiablity mantra there is no reason for a "do-everything" config file [09:48] Justin Clark-Casey: mic: do you know if that code is actually available anywhere else. If not, then publishing it would be great [09:48] Mic Bowman: we have three or for regions currently using extsim [09:48] Mic Bowman: extsim? [09:48] Justin Clark-Casey: yeah [09:48] Mic Bowman: i don't know if its available anywhere other than forge [09:49] Mic Bowman: and in some of the code we've seen from our regions [09:49] Justin Clark-Casey: Oh I see, get it direct from svn I guess then [09:49] Justin Clark-Casey: I see the last change was 3 months ago [09:49] Mic Bowman: the biggest problem is that we have to keep fixing it with changes to the region module changes in opensim [09:49] Mic Bowman: (STABLE!) [09:50] Justin Clark-Casey: so does the code on the forge work? [09:50] Justin Clark-Casey: with the latest OpenSim? [09:50] Mic Bowman: i don't know [09:50] Mic Bowman: i know that the three groups using it have had to hack it recently [09:50] Justin Clark-Casey: ah,. so your changes are still private? [09:50] Mic Bowman: i don't think 3di is working on it [09:50] Justin Clark-Casey: and they didn't republish the changes? [09:51] Justin Clark-Casey: ping [09:51] Mic Bowman: no [09:51] Justin Clark-Casey: ok [09:52] Mic Bowman: and thats why i was thinking that it might be a good idea to grab it [09:52] Mic Bowman: if one goal of the foundation is to create a distribution that makes it easier to do viz [09:52] Justin Clark-Casey: in that case, it definitely would be :) [09:52] Mic Bowman: then that might be exactly the kind of package we should pull in [09:52] Justin Clark-Casey: already sounds like something that many people would contribute to [09:52] Justin Clark-Casey: maybe including me when I need it (and once I've taken a look) [09:53] Mic Bowman: yeah... [09:53] Mic Bowman: i'll talk to the guys at northwestern who are the most recent ones to work on it [09:54] Mic Bowman: rob: could you give me some more information on your memory leak? [09:55] Mic Bowman: how much memory is leaked? how fast? what kind of machine & load on the machine? [09:55] Lightswitch: lights off [09:55] Lightswitch: lights on [09:55] Justin Clark-Casey spills his tea [09:55] Mic Bowman: that was the "wakeup" for rob [09:55] Mic Bowman: :-) [09:55] Justin Clark-Casey: maybe there should be individual spotlights over the seats :) [09:56] Mic Bowman: oh i like that [09:56] Mic Bowman: on a totally unrelated topic... [09:56] Justin Clark-Casey: you'd hit the lightsources limit pretty quickly though [09:56] Mic Bowman: i was looking at penrose tiles like night [09:56] Mic Bowman: last [09:56] Robert Adams: I turned on the lights because I'm running Kirsten's latest shadow viewer and it's pretty dark in here with the sun shining outside [09:56] Mic Bowman: any one know if there is a script for SL to generate quasicrystals or penrose tiles? [09:56] Doug Osborn: cool ting calcs [09:57] Doug Osborn: tiling [09:57] Mic Bowman: that's the non-repeating symmetric patterns [09:57] Doug Osborn: i've not found one [09:57] Mic Bowman: the tile floor has gotten me thinking too much about stuff like that [09:57] Mic Bowman: i have a couple java implementations that i've found [09:57] Doug Osborn: hmm [09:58] Mic Bowman: a simple penrose tile floor should be reasonably easy to make [09:58] Mic Bowman: its just two prim shapes laid down with the appropriate set of rules [09:58] Doug Osborn: thats does sound interesting [09:59] Aaron Duffy: and then once you have the pattern, get a cellular automata changing the colors... [09:59] Mic Bowman: i like the way you think! [09:59] Doug Osborn: :) [09:59] Robert Adams: we have >100K prims.... we now go for >100K scripts [09:59] Mic Bowman: doug: your boss has seriously corrupted my thinking about content in opensim [10:00] Doug Osborn: yes [10:00] Doug Osborn: me too [10:00] Mic Bowman: yeah... that would be a cool demo... 100K prim penrose tile pattern [10:00] Aaron Duffy: can you tile in 3 dimensions? [10:00] Doug Osborn: my bigger probelm is that my appetitie exptends to Sl wher the price per pim is much higher ;( [10:00] Mic Bowman: i didn't find anyone who was working on it [10:01] Mic Bowman: however... the penrose concept is really a projection of five dimensional shapes into 2D [10:01] Mic Bowman: but the math starts to hurt my head [10:01] Aaron Duffy: I've been trying not to distract myself by thinging about setting up an automaton on a 3D grid [10:02] Mic Bowman: i know... [10:02] Mic Bowman: i keep trying to stay focused on the performance issues... but my weekends are spent thinking about apps [10:03] Aaron Duffy: Maybe I can give up sleeping... [10:03] Mic Bowman: at any rate... we have five minutes left... any other topcics? [10:03] Mic Bowman: :-) [10:03] Doug Osborn: im good [10:04] Mic Bowman: great... [10:04] Mic Bowman: i'll post the minutes on the wiki