[09:01] Rob Knop: Morning [09:01] Aaron Duffy: Hi Robert [09:01] Aaron Duffy: It was a meeting of Robs for a minute there [09:01] Rob Knop: It is! [09:01] Robert Adams: or maybe, appropriate-time-of-day-greeting all [09:01] Mic Bowman: good morning everybody [09:02] Rob Knop: Reminds me of a dorm at Harvey Mudd College -- on year, that dorm had more Daves than women. [09:02] Justin Clark-Casey is Online [09:02] Aaron Duffy: :> [09:02] Fei Yeh: :) [09:03] Doug Osborn is Online [09:04] Justin Clark-Casey: hello gentlepeople [09:04] Rob Willis: morning [09:04] Rob Knop: Howdy [09:04] Fei Yeh: hi [09:05] Mic Bowman: be right there... capturing some discussions from the senate meeting re: the foundation [09:05] Doug Osborn: aloha all [09:05] Robert Adams: morning Doug [09:05] Aaron Duffy: hi [09:07] Mic Bowman: would appreciate it if all of you at some point could take a look through http://www.sciencesim.com/wiki/doku.php/foundation/start [09:07] Doug Osborn: robert.... i picked up the binaries for Cable beach early this week..... [09:07] Mic Bowman: for the techies... please look through the development procedures & notes on "stability' [09:08] Doug Osborn: :) [09:08] Mic Bowman: on that note... [09:08] Mic Bowman: i'm in the process of setting up some additional public branches in our distribution [09:09] Rob Willis: the discussions are still being posted? [09:09] Mic Bowman: scisim-stable continues to be the "stable" branch that doesn't change without prior announcements [09:09] Doug Osborn: i like the test cases Mic :) that helps a lot :) [09:10] Mic Bowman: scisim-experimental will be a bleeding edge distribution that is closer to opensim HEAD [09:10] Rob Knop: Re: bi-annual feature releases -- if this sucks in new stuff from core OpenSim, it could well include API changes thatw ill break folks' applications [09:10] Mic Bowman: it generally includes a set of cherry picked commits [09:10] Rob Knop: There should be thought about how long to keep doing bugfixes on a previous feature release [09:10] Rob Knop: For people who don't want to upgrade to the new feature release [09:10] Rob Knop: Perhaps "only security bugfixes" or some such [09:10] Doug Osborn: can we get sandbox regions running each of those....so we can do bug reporting without having to trapse over to different grids? [09:10] Mic Bowman: rob: the bi-annual releases *must* document api changes [09:11] Rob Knop: Right... but even if documented, some people aren't going to want to upgrade if the api changed. [09:11] Mic Bowman: and must be available long before any kind of cut over would occur [09:11] Mic Bowman: that's true... and i think that discussion would be *THE* topic for the advisory board [09:12] Mic Bowman: i need to put together a picture to explain how i see the releases working [09:12] Primitive: Permissions [09:12] Primitive: Taking control of 268436275 [09:12] Rob Knop: E.g., somebody may be using a third party application, and they can't get the third party to update the application, so they're stuck. [09:12] Mic Bowman: that's right... [09:12] Doug Osborn: they wouldnt be stuck if they documented the api as a requirement [09:12] Mic Bowman: six months is far more stability than you get right now on APIs... but it isn't long enough [09:13] Mic Bowman: i think the real hope is that projects like VWRAP will start to provide standards-based APIs [09:13] Mic Bowman: with much broader reach [09:13] Fei Yeh is Online [09:13] Mic Bowman: anyway... do you think six months is too fast? what would you recommend? [09:14] Doug Osborn: It going to depend a lot on the speed at which core puts out updates as well.... [09:14] Rob Knop: Yeah... I don't have a real sense about VWRAP. i have to admit that my perception is that Linden and the whole OGP/VRAP thing kind of fell by the wayside about a year ago, and that OpenSim moved on (with things like HyperGrid). How do the OpenSim core developers view VWRAP? Are they actively participating? [09:14] Mic Bowman: there is some active participating [09:14] Justin Clark-Casey: a few people follow the mailing lists but activity is low at the moment [09:14] Mic Bowman: the cable beach project we use here on sciencesim is much closer to vwrap [09:14] Mic Bowman: so there are "real" implementations [09:15] Doug Osborn: I think 6 months is a good timetable to start ....given the manpower needs to do it more rapidly might not be available in the sttup phases of the projects too [09:15] Doug Osborn: *startup [09:15] Rob Knop: There will also need to be two test infrastructures -- the next feature release will probably need a two-month test cycle (or some period much longer than a week, depending on how many people are paid and how many volunteers contribute) [09:15] Rob Knop: That should be separated from the bugfix testing so they don't fight for resources. [09:15] Mic Bowman: that's correct [09:15] Rob Knop: Hmmm [09:16] Rob Knop: A long line of chat I typed never showed up [09:16] Mic Bowman: what i had in mind was actually three [09:16] Mic Bowman: 1) based on the stable release and provides public core services [09:16] Rob Knop: Or did it? Did you see the long line where I said that the feature release may take two months of testing? [09:16] Rob Knop: I didn't see that echoed back to me [09:16] Doug Osborn: i see your chat [09:16] Mic Bowman: 2) for testing and debugging the next bi-weekly release [09:16] Rob Knop: Just that one line is what I missed [09:17] Mic Bowman: 3) for experimentation with the new feature releases [09:17] Justin Clark-Casey: all this depends on exactly 'how much' stability you want. There can be very hard to pin down bugs that may take longer [09:17] Mic Bowman: "stable" isn't "yes" or "no" [09:17] Rob Knop: Yeah... it will take judgement on the part of whoever is in charge of release and QA [09:17] Mic Bowman: it has to be based on a set of tests [09:18] Mic Bowman: jcc: did you see the "stable" notes i put up the other day? a bit more esoteric than i want... but its a start [09:18] Mic Bowman: http://www.sciencesim.com/wiki/doku.php/foundation/stable [09:18] Doug Osborn: yes I would very much like to have a test suite of scripts [09:19] Mic Bowman: basic idea is to define stable in a very operational way... based on execution of a series of tests [09:19] Justin Clark-Casey: no, this is the first time I've followed this link [09:19] Mic Bowman: not just unit tests, but also "bot driven" and "human driven" test cases [09:19] Rob Knop: Yeah, human driven test cases are important [09:19] Mic Bowman: and it includes performance tests [09:20] Mic Bowman: i wasthinking that we (intel) could contribute a bunch of our automated benchmark tests as one place to start [09:20] Rob Knop: But I'd also [09:20] Rob Knop: d'oh [09:20] Rob Knop: I'd also include judgement as part of the tests. Things may break in a way that nobody thought to create a test for [09:20] Iti Sugarplum: hello Mic [09:20] Mic Bowman: hi iti [09:20] Rob Knop: You do say that tests will evolve. [09:20] Mic Bowman: agreed rob... [09:20] Iti Sugarplum: I am lost [09:20] Rob Knop: But it may be possible that a release passes all the tests that are indicated for it, but is still "broken". [09:20] Justin Clark-Casey: The API driven note at the end may be difficult to maintain in the face of changes in the base OpenSim [09:20] Iti Sugarplum: i want to go to the place where there is a big pink stree [09:21] Rob Knop saw that a lot at Linden when he was server release manager.... [09:21] Mic Bowman: what i was thinking was that the "output" of the tests be automated & documented as part of that [09:21] Doug Osborn: yes thats why its important for people who rely on third party applications to come forward .....maybe the interface to the application can be designed as a test case.... [09:21] Iti Sugarplum: sorry to interrupt you [09:21] Rob Knop: JustinCC -- is there any drive amongst the OpenSim core developers to stabilize the API at all, or is that not on the radar yet? [09:21] Mic Bowman: iti: big pink tree? [09:21] Iti Sugarplum: yes [09:21] Iti Sugarplum: i am buidling something there [09:21] Iti Sugarplum: i forgot to take landmark down [09:21] Mic Bowman: oh... that's just west of here [09:22] Iti Sugarplum: big pink tree [09:22] Justin Clark-Casey: rob: there's no formal or consistent drive yet [09:22] Iti Sugarplum: thanks [09:22] Rob Knop: Hmm... when i type a long line of chat, it doesn't seem to come back to me right away, but basd on responses I see that others have seen it. [09:22] Doug Osborn: im seeing the same chat issue rob [09:22] Mic Bowman: sigh... [09:22] Rob Knop: The hardest bugs are the intermittent ones.... [09:23] Justin Clark-Casey: it's difficult to stabilize a project where even chat isn't working properly ;) [09:23] Doug Osborn: but only when i use the chat box...if u se the local chat line at the bottom of the screen everything pops up right away [09:23] Mic Bowman: jcc: why? [09:23] Rob Knop: Interesting [09:23] Justin Clark-Casey: poits to the general immaturity of the codebase, I think [09:23] Mic Bowman: if we focus on fixing the bugs (like chat) rather than yet another refactor... [09:23] Rob Knop: There's probably enough experience now, though, that a lot of the API itself could be stabillized... even if there are still functionality problems in parts of it. [09:24] Doug Osborn: stable doesnmt mean matre...it means ditroas dont inlcude fixes to bugs that introduce new ones to existing features :) [09:24] Mic Bowman: the code base is no more immature than many shipping products [09:24] Justin Clark-Casey: true, but bug fixing isn't api stabiliziation [09:24] Rob Knop: At least, a version of the APi could be stabilized; VWRAP may throw a wrench into a lot of it. [09:24] Mic Bowman: the immaturity is more because of lack of focus (mic's opinion...) [09:25] Mic Bowman: rob: like i said... the CB architecture that we're already running here is just a delta off what vwrap is proposing [09:25] Justin Clark-Casey: yeah, I do honestly think that's a fair point. It requires people with a strong reason to make it stable [09:25] Justin Clark-Casey: relatively conservative types :) [09:25] Doug Osborn: and the API argument becimes much less esoteric when there is a working model that provides the required features... a working Cable beach is easier to understand and move on then a "possible VWrap API that looks pretty [09:25] Rob Knop: Mic : how large a project is Cable Beach? That is, how much of the OpenSim fucntionatliy goes through Cable Beach interfaces and such? [09:26] Mic Bowman: most of cable beach is currently implemented as region modules for the simulator, service modules for asset & inventory and big patch for the user server [09:26] Mic Bowman: the code is all in the scisim source repository in the add-ons directory [09:26] Rob Knop: (Is there a Cable Beach "home page" or overview?) [09:27] Rob Knop: Ah, OK [09:27] Mic Bowman: just a second... [09:27] Doug Osborn: and the binary distros are very nice :) [09:27] Mic Bowman: http://code.google.com/p/cablebeach/ [09:27] Mic Bowman: robert adams: when did we last update the binary distribution? [09:28] Iti Sugarplum is Online [09:28] Lightswitch: lights on [09:28] Lightswitch: lights off [09:28] Mic Bowman: see if that works to wake up robert [09:28] Mic Bowman: :-) [09:28] Robert Adams: serveral weeks ago [09:28] Doug Osborn: yes I would request a note on the wiki where the distros are posted with a OpenSim base release... [09:28] Rob Knop: Is there any thought/hope that Cable Beach may eventually get integrated into OpenSim core? [09:28] Mic Bowman: doug: you're right... [09:28] Mic Bowman: rob: that's a policy decision [09:28] Doug Osborn: rib : all the patches are submitted to core [09:29] Mic Bowman: philopsophically... i'm against putting it in core if it can be implemented as modules [09:29] Mic Bowman: practically... if it simplifies implementation & distribution i'm not against it [09:29] Justin Clark-Casey: it was in core before but nobody maintained it, so it was removed again [09:29] Rob Knop: Well, true... I guess what I mean is, "become part of the OpenSim standard that 'everybody uses'" [09:30] Doug Osborn: Its not the right solution for everyone [09:30] Mic Bowman: well... there would be a certain upheaval with the hypergrid 2 folks [09:30] Mic Bowman: because the two are similar... but not really compatible [09:30] Rob Knop: Ah.... I didn't realize Cable Beach was orthogonal to HyperGrid 2 [09:30] Mic Bowman: orthgonal is too strong [09:30] Rob Knop: Cable Beach includes the cability to move between grids? [09:30] Mic Bowman: they are solving similar problems with a different trust model [09:31] Doug Osborn: HG2 doesnt have a way to maintain identity credaetioals that are importatn to the design of assett access in Cable Beach [09:31] Mic Bowman: it includes the ability to have your own identity and inventory among many grids [09:31] Rob Knop: Hmm... for something like interoperability, eventually there's going to need to be a dominant standard. I guess it's too soon to know which it wil be? [09:31] Mic Bowman: like hypergrid... the actual movement between grids can be as transparent as you want the viewer to make it [09:31] Mic Bowman: interoperability is like stability... its a spectrum [09:32] Mic Bowman: vwrap seems to be scoped for success if they can move quickly enough into tangible implementations [09:32] Rob Knop: Yeah... but in the medium/long term, the world's going to want to have a single protocol for moving between grids. (Nobody uses Novell Netoworks anymore, everythign is TCP/IP, etc.) [09:32] Mic Bowman: far better than the "world of warcraft to second life" discussions [09:32] Rob Knop: Heh! [09:33] Doug Osborn: Rob: you need to explain that to the people who pay me to fix thier broken Nvell server implements :) [09:33] Rob Knop can't imagine WoW *really* being relevant to the long-term future of the metaverse... [09:33] Mic Bowman: cable beach basically allows you to take your identity & inventory... if we had a viewer to which we could give a loginuri while it was running, then we would have grid to grid with secure access to inventory [09:33] Rob Knop: Doug : people *do* still use it??? wow. Are there still new installations going up, or is this legacy stuff that people want to keep using without redoing their net? [09:33] Justin Clark-Casey: Isn't think thinking too long term? Isn't the first step to try and produce a basic system that is halfway stable? [09:34] Mic Bowman: jcc: yes [09:34] Rob Knop: Yeah, I'm getting ahead of the game, sorry. [09:34] Doug Osborn: Rob Legacy stuff for local school districts who thought "free" included maintenace..... [09:34] Mic Bowman: and to get systems that are running off the same code base to interoperate [09:34] Rob Knop: Doug : d'oh. [09:35] Doug Osborn: JCC: thinking ahead can save developer time in the short run too.... if you can decide on a path you can triage the bug work that isnt in that direction [09:35] Mic Bowman: jcc: i think the highest priority is to establish the scope [09:36] Mic Bowman: what has to work... and what is allowed to break [09:36] Justin Clark-Casey: doug: there's medium term nad then there's long term :) [09:36] Justin Clark-Casey: mic: isn't that more or less going to be windowdressing right now? Plans will 100% certainly change ove rthe long term. [09:36] Mic Bowman: write the tests for what has to work... and fix the bugs [09:36] Mic Bowman: absolutely [09:36] Mic Bowman: but that's what product development is about [09:36] Justin Clark-Casey: better off throwing something reasonably convincing up and then trying to meet short term objectives of stability and community building, imho :) [09:37] Justin Clark-Casey: well, absolutely [09:37] Mic Bowman: what do you want to build [09:37] Doug Osborn: take libraries... the IAR library discussion has led to a new interst there.... but a simple decision to adopt a "real" stabdard like web-DAV to support web access to asset libraries might be a good thing no??? [09:37] Mic Bowman: if you want something that slices & dices and toast your bread... that's going to be tough [09:37] Mic Bowman: but if you want a knife sharpener... thats easier [09:37] Justin Clark-Casey: yeah, but takes longer to develop. IAR is a low cost interim solution, and those are often the best ones [09:38] Justin Clark-Casey: you can have 1000 discussions about remote asset access but I think that one has to climb the low hill first [09:38] Mic Bowman: this from the guy who's building the slice/dice/toast platform???? :-) [09:38] Justin Clark-Casey: and indeed, much remote access discussion has been had in the past without any code to show for it [09:38] Justin Clark-Casey: well, maybe I am, maybe I'm not, who knows :) [09:38] Mic Bowman: jcc: john's schedule for webdav and cross grid inventory is end of next week [09:39] Doug Osborn: jcc: agreed but there are imprtant discussion in the IAR stuff that are still not addressed.... we are stil dancing around creator IDS and how they get stored... [09:39] Justin Clark-Casey: excellent - I'll loko forwad to the revolution [09:39] Justin Clark-Casey: doug: that's a really hard problem [09:39] Mic Bowman: :-) [09:39] Justin Clark-Casey: in the context of the current architecture [09:39] Mic Bowman: i'll be happy if it just works... [09:39] Rob Knop: cd [09:39] Rob Knop: (sorry, wrong window) [09:39] Mic Bowman: linux guys... [09:40] Doug Osborn: if you see IAR as a step up the path to web access you acknowldge the need to address creator tags and can move on it.... it doesnt make the IAR any less useful... but it can be included in a stable design API design [09:40] Mic Bowman: can we change subjects for a minute... [09:40] Mic Bowman: some thing far more tactical [09:40] Justin Clark-Casey: iar is a ghetto format, but unlike a lot of the discussion it actually exists [09:40] Doug Osborn: and works :) i made my first one last night [09:40] Mic Bowman: we've been having a number of problems with mysql connections being dropped and not recreated [09:41] Mic Bowman: john patched the mysql code to open the mysql connections on each request [09:41] Mic Bowman: with connection pooling, that should still perform well [09:42] Mic Bowman: and it avoids many of the problems we've been having with large numbers of persistent connections to the database server [09:42] Mic Bowman: i'm going to merge that code into the scisim experimental branch later today if you'd like to test it [09:43] Mic Bowman: we've done a number of "nasty" load oar tests to push the region db into unnatural acts and the db has not crashed [09:43] Mic Bowman: which is an improvement... [09:43] Mic Bowman: but the code needs a lot more testing [09:44] Mic Bowman: so if you're interested... pull the new repository and test it out... but back up your DB first [09:44] Mic Bowman: experimental also has a number of additional bug fix commits from the opensim tree [09:44] Mic Bowman: doug: you recommended documentation... do you have a suggestion for what you want to see? [09:45] Mic Bowman: is it enough to do a git log --pretty=full [09:45] Mic Bowman: and dump that to a web page somewhere? [09:46] Doug Osborn: well we are working with Adam on the V-business grid and trying to get an "adminstrators crib sheet out of him on the latest configuration op[tions [09:47] Doug Osborn: I think its important to make the install as pain free as possible so we can have as many people as possible close to the front of the product [09:47] Mic Bowman: you're talking about documenting administration? [09:48] Doug Osborn: i think there wwas some discussion on the OPensim Wiki or mailing list about segregating platforms and release specific stuff... It would be nice if we could implement something like that for CB... so we dont have to review all of the wiki pages when a fix for a windows version comes out [09:49] Doug Osborn: and people dont see notes about stuff from 8 versions ago [09:49] Rob Knop: I need to go -- have to move to SL for Shenlei's talk, and (of course) I'm having trouble getting Voice working again [09:49] Mic Bowman: if i give you a namespace on the wiki, would you put together a skeleton of the pages you'd like to see? [09:50] Mic Bowman: rob: thanks for stopping by [09:50] Rob Knop is Offline [09:50] Doug Osborn: I was also talking to P.Finn last night about whether he had anyone at IBM looking at software flow visualization... maybe we need a couple regions somewhere with 3d flowcharts or design documents so people can see the structure of some/all of the Opensim design... [09:50] Doug Osborn: i know you love visualization projects here :) [09:50] Mic Bowman: what's the status of the 3d wiki work? [09:51] Doug Osborn: you have to ask shen about that...she's kinda taken that project [09:52] Doug Osborn: and yes Id be happy to try and flesh out some wiki pages [09:53] Mic Bowman: http://www.sciencesim.com/wiki/doku.php/admin/start [09:53] Doug Osborn: got it [09:54] Mic Bowman: i need to take off for now... [09:54] Mic Bowman: thanks for the great discussion this morning [09:54] Doug Osborn: good to see you all... :) [09:54] Justin Clark-Casey: I'm going too. Goodbye folks [09:54] Justin Clark-Casey is Offline [09:55] Doug Osborn: robert... I installed the linux binary and seem to have an error in the console when i ask for a show regions????? [09:55] Doug Osborn: everything esle seems to be working fine..... [09:55] Fei Yeh: yeah that error's been there for a while [09:55] Mic Bowman: would you post that to either scisim-support or scisim-discuss [09:55] Robert Adams: that error has been around for a while [09:55] Doug Osborn: OK [09:56] Mic Bowman: i need to patch some stuff in that file anyway [09:56] Doug Osborn: is it mantised? [09:56] Mic Bowman: and should just take a look