Chat log from the meeting on 2009-11-17

From OpenSimulator

Jump to: navigation, search
Tuesday Meeting @ Wright Plaza on OSGrid


[10:54] Adelle Fitzgerald: hiya teravus
[10:54] Teravus Ousley: hi
[10:55] Teravus Ousley: everything is still a bit halloween-ish here.
[10:55] Adelle Fitzgerald: hehe yeah
[10:55] Adelle Fitzgerald: could do with putting back to normal now i think
[10:56] Adelle Fitzgerald: i cant say i will miss the bird noises
[10:56] Teravus Ousley: heh
[10:56] Verlian Shambala is Online
[10:56] Teravus Ousley thinks he messed up the meeting time or something
[10:57] Adelle Fitzgerald: hello Long, or is it just John :)
[10:57] Adelle Fitzgerald: hi Melanie
[10:57] Long John: hi:)
[10:57] Long John: Long or John
[10:58] Adelle Fitzgerald: kk
[10:58] Verlian Shambala is Offline
[10:59] Justin Clark-Casey is Online
[10:59] Melanie Milland: hi
[10:59] Teravus Ousley: hello
[10:59] Adelle Fitzgerald: hi Penny
[10:59] Justin Clark-Casey: hello folks
[10:59] Snoopy Pfeffer is Online
[10:59] Adelle Fitzgerald: hello
[10:59] Teravus Ousley: hello Justincc
[10:59] Friendly Harbour: hello everyone
[11:00] Nalani Moleno: hello everybody!
[11:00] Teravus Ousley: I saw your chatter on the SC conference chat
[11:00] Snoopy Pfeffer: hello all :)
[11:00] Penny Lane: Hi Adelle :-)
[11:00] Penny Lane: Hi all :-)
[11:00] Justin Clark-Casey: Verbal diahrrea. That thing became an OpenSim chatroom
[11:00] Teravus Ousley: It did, indeed. Even when they were talking about the Larabee
[11:01] Gigadreamer Bazar is Online
[11:01] Justin Clark-Casey: yeah, hope that didn't annoy other people too much
[11:01] Snoopy Pfeffer: did I miss something? :)
[11:01] Melanie Milland: well, the message that opensim is the next apache was loud and clear
[11:01] Snoopy Pfeffer: do you talk about irc?
[11:01] Teravus Ousley: It certainly was there.
[11:02] Justin Clark-Casey: yes. I think it remains to be seen what effect it will have - depends on whether it gels with the audience I think
[11:02] Teravus Ousley: There was a keynote by Intel's Justin Rattner (CTO) today at the Super Computing conference.
[11:02] Snoopy Pfeffer: oh cool
[11:03] Teravus Ousley: it mentioned OpenSimulator, ScienceSim, the Fashion Research Institute, Bullet Physics.. and other cool related technologies
[11:03] Snoopy Pfeffer: great
[11:03] Friendly Harbour: sorry i missed it
[11:04] Friendly Harbour: is it available online somewhere now?
[11:04] Teravus Ousley: Diva also gave a small presentation there on OpenSimulator.
[11:04] Snoopy Pfeffer: is that somewhere on the internet?
[11:04] Snoopy Pfeffer: wow cool
[11:04] Teravus Ousley: It was recorded, but isn't currently available.
[11:04] Friendly Harbour: ok
[11:04] Snoopy Pfeffer: ok
[11:04] Penny Lane: They'll be putting the video up late, I heard said.
[11:04] Friendly Harbour: great
[11:04] Penny Lane: later*
[11:04] Justin Clark-Casey: snoppy: there should be an update on twitter.com/isntv if it does go up
[11:05] Snoopy Pfeffer: ok
[11:05] Justin Clark-Casey: some concerns about permission to rebroadcast I seem to remember
[11:05] Snoopy Pfeffer: hm always there are concerns about permissions ;) lol
[11:05] Teravus Ousley: So, how are we doing today in OpenSimulator?
[11:05] Snoopy Pfeffer: everywhere ;)
[11:05] Justin Clark-Casey: teravus: heh, not being able to land on prims :)
[11:06] Teravus Ousley: Yes, that's a difficult issue to solve, as I mentioned previously
[11:06] Snoopy Pfeffer: some of my renters like to build vehicles
[11:06] Snoopy Pfeffer: that llSetPos and llSetRot bug is really annoying for them
[11:06] Justin Clark-Casey: previously in irc?
[11:06] Snoopy Pfeffer: yes
[11:07] Snoopy Pfeffer: llSetPos was discussed
[11:07] Snoopy Pfeffer: llSetRot*
[11:07] Snoopy Pfeffer: llSetPos is the same
[11:07] Teravus Ousley: Yes to mine also. in IRC and on my blog page.
[11:07] Snoopy Pfeffer: the object moves, but nobody sees the movement
[11:07] Justin Clark-Casey: teravus: what is your blog address?
[11:07] Teravus Ousley: b50d.com
[11:07] Justin Clark-Casey: snoopy: sounds like an update is missing
[11:08] Snoopy Pfeffer: yes
[11:08] Snoopy Pfeffer: that was discussed 2 weeks ago
[11:08] Gigadreamer Bazar is Offline
[11:08] Snoopy Pfeffer: that rotation problem
[11:08] Teravus Ousley: have a look at the cube in the middle of the moon ground.
[11:08] Snoopy Pfeffer: was there a bug fix in the meantime?
[11:08] Teravus Ousley: I'll demonstrate the issue there.
[11:08] Hiro Protagonist is Online
[11:08] Snoopy Pfeffer: oh that shifting
[11:08] Teravus Ousley: see it floating away?
[11:08] Justin Clark-Casey: teravus: I get a server not found with that url
[11:09] Snoopy Pfeffer: I also experience tat with avatars!
[11:09] Snoopy Pfeffer: that*
[11:09] Snoopy Pfeffer: sometimes I start drifting
[11:09] Teravus Ousley: well, it turns out that once that happens, the object and physics are no longer in sync
[11:10] Teravus Ousley: so, if I stood up and hit the object with my avatar, it would no longer collide properly
[11:10] Teravus Ousley: .. physics wise, it would move
[11:10] Adelle Fitzgerald: hello Hiro
[11:10] Teravus Ousley: but there's some disconnect there
[11:10] Hiro Protagonist: Hey everbodeh
[11:10] Snoopy Pfeffer: hi Hiro
[11:10] Teravus Ousley: and it never bubbles to the client
[11:11] BlueWall Slade is Online
[11:11] Adelle Fitzgerald: ive noticed of late that a region physics FPS, sim FPS and TD all just seem to stop. the region doesnt recover and a lot of the time you have to completely terminate the process. shutting it down clean doesnt always work to restart it
[11:12] Justin Clark-Casey: teravus: so that means the physics represenation of a floor gradually slides away over time?
[11:12] Adelle Fitzgerald: i havent been able to pin it on anything other than high traffic regions seem to be more affected
[11:12] Teravus Ousley: no, it means that somewhere between physics, and the update to the client, there's a disconnect.
[11:13] Justin Clark-Casey: so how does this relate to the problem of not being able to land on prims automatically any more?
[11:13] Teravus Ousley: I checked in physics, and physics positions, rotations, linear velocity etc.. are ok
[11:13] Teravus Ousley: Landing requires a specific set of conditions to exist.
[11:14] Teravus Ousley: certain velocities and positions along with certain buttons being pressed
[11:15] Justin Clark-Casey: and it's difficult to recreate what we had before?
[11:15] Teravus Ousley: On the simulator, it recognizes these situations, but the updates to the client do not trigger the client to flip the fly bit because the situation no longer matches what the client will flip the fly bit on.
[11:15] Dahlia Trimble is Online
[11:15] Hiro Protagonist is Online
[11:16] Teravus Ousley: It was difficult to get it to work previously. It'll be difficult to get it to work again.
[11:16] Justin Clark-Casey: oh god
[11:16] Dahlia Trimble: hi :)
[11:16] Justin Clark-Casey: hello dahlia
[11:16] Andress Renault: Yes M'child..sorry could not resist.
[11:16] Justin Clark-Casey: ha.
[11:17] Snoopy Pfeffer: Teravus, in mantis #4336 you mention changes in ODE (PhysicsVector replaced by Vector3) - are there many changes that it requires to check the whole ODE integration?
[11:17] Teravus Ousley: I'm sure with perfect information, it would be easier. But, since we don't have perfect information, it is what it is :)
[11:17] Teravus Ousley: Snoopy, I've looked and the ODE->ODEPlugin layer seems to be okay.
[11:17] Snoopy Pfeffer: ah ok thanks
[11:18] Teravus Ousley: http://b5od.com
[11:18] Snoopy Pfeffer: so probably someone simply broke sending object position / rotation updates
[11:18] Hiro Protagonist: hot topic of late
[11:19] Hiro Protagonist: anyone seen Nebadon today?
[11:19] Teravus Ousley: some portion of the layer between the ODEPlugin--->Client update has a disconnect.
[11:19] Snoopy Pfeffer: ok
[11:19] Snoopy Pfeffer: so somewhere there is the reason for that llSetPosllSetRot bug
[11:19] Justin Clark-Casey: teravus: ah a letter o not a zero. Now I just get Bad Request (Invalid Hostname)
[11:19] Justin Clark-Casey: teravus: So presumably this is something to take up with jhurliman?
[11:20] Teravus Ousley: it requires that the browser send a host header :)
[11:20] Nebadon Izumi is Online
[11:20] Teravus Ousley: It's really something that we should all be looking at and trying to figure out.
[11:20] Justin Clark-Casey: got it :)
[11:20] Hiro Protagonist: Here's Neb :)
[11:20] Nebadon Izumi: hello
[11:20] Snoopy Pfeffer: hi Neb
[11:20] Justin Clark-Casey: either way, it's regression-city
[11:20] Justin Clark-Casey: hi neb
[11:21] Andress Renault: Hey Neb :)
[11:21] Teravus Ousley: It's definately frustrating to see it occur, but it isn't unsolvable.
[11:21] Hiro Protagonist: we were just talking about some new things Teravus is seeing and whether they might be associated with the llSetPos/llSetRot bug
[11:21] Nebadon Izumi: sorry im late, feel really tired today.. not getting good sleep lately hehe
[11:21] Teravus Ousley: it will just take a good concerted effort to resolve each regression.
[11:22] Justin Clark-Casey: teravus: I find it frustrating as well, esp. as it will take me enormously longer to come anywhere near up to speed on this stuff
[11:22] Teravus Ousley: It means donating time to solve the issues and produce fabulous bug reports that help developers get to the root of the issue.
[11:22] Maria Korolov is Online
[11:22] Dahlia Trimble: best way I've found so far to resolve regressions is to do a binary search through past revisions and find the offending commit
[11:23] Nebadon Izumi: i gotta say i find that extremly difficult with git
[11:23] Teravus Ousley: I havn't mastered git in that regard.
[11:23] Warin Cascabel is Online
[11:23] Nebadon Izumi: if not near impossible
[11:23] Dahlia Trimble: which was easier when we had SVN
[11:23] Justin Clark-Casey: best way I've found is to write an automated test to make sure it don't happen again :)
[11:23] Nebadon Izumi: honestly i dont have the time to do it in git
[11:23] Hiro Protagonist: Here's what I know about the problem
[11:23] Teravus Ousley: I agree justincc.
[11:23] Andress Renault listens.
[11:23] Teravus Ousley: that's probably the /best/ way to assure it doesn't happen again.
[11:24] Dahlia Trimble: automated tests havent helped me on OpenSim yet :/
[11:24] Hiro Protagonist: Certain people blame it on work by JHurliman. That in itself isnt really constructive, but the timing might be usefull.
[11:24] Justin Clark-Casey: but time consuming I know
[11:24] Snoopy Pfeffer: doesnt the git log show the sequential order or the commits and their keys?
[11:24] Hiro Protagonist: Other people (Melanie_T) actually took the responsibility for the problem
[11:24] Hiro Protagonist: I think she can probably say exactly what commit it was that introduced the issue
[11:24] Penny Lane: It's the only way. Test scripts scale. You guys don't, you need sleep :P
[11:24] Nebadon Izumi: snoopy there is no real way to easily look at the git log
[11:24] Justin Clark-Casey: there's no point in spreading blame - it's just something that needs to be worked through
[11:24] Nebadon Izumi: and move fast through revisions
[11:25] Snoopy Pfeffer: Neb, ok
[11:25] Nebadon Izumi: its very time consuming
[11:25] Nebadon Izumi: and frustrating
[11:25] Teravus Ousley: It doesn't really help to blame people. With each commit something changes. Sometimes for better, some times for worse, sometimes both. It's okay and normal in a project's development process.
[11:25] Dahlia Trimble: theres a "git gui" that helps
[11:25] Hiro Protagonist: I know Justin, that's not my point at all (blaming anyone)
[11:25] Hiro Protagonist: I just think I know where to find out the commit and probably something about the issue
[11:25] Nebadon Izumi: to me the git GUI doesnt make it any easier
[11:25] Nebadon Izumi: when your 50 revisions back
[11:25] Nebadon Izumi: its impossible to step through
[11:25] Hiro Protagonist: CasperW swears he has fixed it
[11:26] Teravus Ousley: jhurliman is an easy target.. because he's rewritten things and done a lot of work on OpenSimulator the past month.
[11:26] Snoopy Pfeffer: I think I saw me drifting the first time about nearly 2 weeks ago
[11:26] Hiro Protagonist: no doubt
[11:26] Nebadon Izumi: i tell you what anyone who claims they fixed something
[11:26] Teravus Ousley: but, that doesn't necessarily make him the cause of the issue. :)
[11:26] Nebadon Izumi: and isnt sharing it
[11:26] Nebadon Izumi: is going to get a nice smack from me
[11:26] BlueWall Slade: ha
[11:26] OtakuMegane Desu: Even if he was, I'd say the positives of his work are worth it anyway.
[11:26] Snoopy Pfeffer: good Neb :)
[11:26] Hiro Protagonist: I dont think it really matters whodunnit except perhaps that it may help isolate where the problem is
[11:26] Andress Renault: Man is wearing a Katana I would not argue.
[11:27] Teravus Ousley: Otaku: that's quite true.
[11:27] Nebadon Izumi: im tired of hearing whining babies blaming devs for stuff
[11:27] Hiro Protagonist: me too
[11:27] Nebadon Izumi: then claiming they fixed it butt refuse to share
[11:27] Nebadon Izumi: cause someone hurt thier feelings
[11:27] Hiro Protagonist: same here
[11:27] OtakuMegane Desu: Lol butthurt
[11:27] Adelle Fitzgerald: does anyone have a recent(ish) git rev# where the bug is not present?
[11:27] Nebadon Izumi: that will seriously earn you a ban on this grid
[11:28] M1sha Dallin is Online
[11:28] Teravus Ousley: There were two merges in quick succession that, before the merges, the issue isn't present.. but a few revisions after the merges, the issue is present.
[11:28] Justin Clark-Casey: I think the landing and drifting issues are really a version holdup. imho we couldn't make a good release with those present
[11:28] Hiro Protagonist: frankly I am amazed that JHurliman and crew could produce such a volume of work and only have one major bug as a side effect\
[11:28] Snoopy Pfeffer: when was that? do you have a date?
[11:28] Hiro Protagonist: they have my full endorsement and admiration
[11:28] WhiteStar Magic: +1 Hiro !
[11:28] Hiro Protagonist: that is, of course, assuming the bug comes from that quarter at all
[11:29] Teravus Ousley: I say a few after the merges because some of the revisions after the merges are untestable because they freeze immediately
[11:29] Dahlia Trimble: which bug are we talking about?
[11:29] BlueWall Slade: I have 71c9291 that updates prims using llSetPos properly
[11:29] Andress Renault: Are you referring to the one where you start sliding across the ground?
[11:29] BlueWall Slade: it doesn't dispose primsthough
[11:30] Teravus Ousley: It's hard with git to pick a 'revision' or to map out the commit structure in a way that's useful in finding where something went wrong.
[11:30] Snoopy Pfeffer: I had that problem with 82554e (2nd Nov) the first time
[11:30] OtakuMegane Desu: Hmm
[11:30] BlueWall Slade: the git bisect can get you on it pretty quick
[11:30] Justin Clark-Casey: Teravus: so you think that isolating the specific revisio nwhere this started wouldn't help?
[11:30] Nebadon Izumi: ya its impossible from my perspective
[11:31] Nebadon Izumi: your talking about potentially days if your talking about cross referencing 100 revisions
[11:31] Adelle Fitzgerald: ok, that rev BlueWall gave is from the 23/10
[11:31] OtakuMegane Desu: This is the problems with updating prim positions and related, right?
[11:31] WhiteStar Magic: 71c9291 = r 11319
[11:31] Adelle Fitzgerald: so that is a start
[11:31] Teravus Ousley: Justin. I don't really because it happened so long ago, and there were so many changes in that one merge that picking one is going to be difficult.
[11:31] Hiro Protagonist: Teravus: Surely it would narrow the field of examination somewhat
[11:32] Snoopy Pfeffer: is was before the 2nd!
[11:32] Nebadon Izumi: we could be talking about 500+ revisions
[11:32] WhiteStar Magic: 82554e - r11432
[11:32] Hiro Protagonist: all the more reason to narrow them down if possible
[11:32] BlueWall Slade: git bisect
[11:32] OtakuMegane Desu: The build I was using from trunk up till a couple days ago was from Oct 27 and llSetPos (and a few other related things) was giving me issues that far back already.
[11:32] Nebadon Izumi: ya im not saying we shouldnt
[11:32] Nebadon Izumi: im just saying its gonna be very time consuming
[11:32] Teravus Ousley: if someone could find the proper commit it would be helpful. But.. finding the merge isn't.
[11:32] Nebadon Izumi: it could take days literally
[11:32] Snoopy Pfeffer: it is all revicions between Oct 27 and Nov 2
[11:33] Snoopy Pfeffer: revisions*
[11:33] Teravus Ousley: Snoopy, that doesn't really help.
[11:33] Teravus Ousley: A merge took place at that time.
[11:33] Hiro Protagonist: yes
[11:33] Snoopy Pfeffer: too bad
[11:33] Hiro Protagonist: a couple in fact
[11:33] BlueWall Slade: the prims started sticking in the scene when we were piling on WP for load teseting
[11:33] Teravus Ousley: The commit that was merged was before that date
[11:33] Justin Clark-Casey: it shouldn't take so long if one properly does divided and conquor. But if a lot was happening at once it may not be that useful, as Teravus says
[11:33] Teravus Ousley: but, it was in a different branch at the time.
[11:34] OtakuMegane Desu: Hmm
[11:34] Nebadon Izumi: the branches that are gone now
[11:34] Nebadon Izumi: are they totally gone?
[11:34] Nebadon Izumi: or can we step back to those still as well?
[11:34] Nebadon Izumi: ive never tried
[11:34] Teravus Ousley: a git person should chime in here :)
[11:34] Penny Lane: Seems odd to me as an observer that you're looking for the faulty commit, instead of actually looking at why it's happening directly through diagnostics :-)
[11:35] Nebadon Izumi: hehe is there any git people here?
[11:35] Dahlia Trimble: there's hopefully a branch revision history, but I havent tried it
[11:35] Nebadon Izumi: i dont think there are
[11:35] Nebadon Izumi: lol
[11:35] Dahlia Trimble: Melanie seems to be the git guru lately
[11:35] Nebadon Izumi: ya i fear without melanie
[11:35] Nebadon Izumi: we are in trouble
[11:35] Nebadon Izumi: hehe
[11:35] Maria Korolov is Online
[11:35] BlueWall Slade: it seems that things started happening arround the time the packet prioritization work started
[11:36] Justin Clark-Casey: someone else would have to learn git properly
[11:36] Nebadon Izumi: hehe
[11:36] BlueWall Slade: but - there were many things going on then
[11:36] Dahlia Trimble: there were several branch merges around that time
[11:36] OtakuMegane Desu: So the closest we can probably pin down so far is whatever commit did it was in trunk Oct 27 or earlier.
[11:36] Nebadon Izumi: since i missed the begining of the meeting
[11:37] Nebadon Izumi: i assum we are talking about the packet stuff?
[11:37] Teravus Ousley: Watch the cube at the middle of the moon.
[11:37] Justin Clark-Casey: nebadon: we're talking about the drifting of physical prims and the fact that we can't automatically land on prims anymore
[11:37] BlueWall Slade: but, tere were troubles before that, because the scene didn't seem to get updated and was leaving prims in after llDie
[11:37] Dahlia Trimble: Is there a specific mantis entry which describes the bug of choice today?
[11:37] OtakuMegane Desu: Ohh yeah. The sinking cubes.
[11:37] Nebadon Izumi: ah
[11:37] Penny Lane: It's like trying to figure out what's wrong with a patient by checking which medication he's been taking, instead of by looking at his body.
[11:37] BlueWall Slade: ohhh, ok
[11:37] Nebadon Izumi: thats been doing that for months
[11:37] Nebadon Izumi: hasnt it
[11:37] Nebadon Izumi: that started way before Jhurliman did anything
[11:37] Snoopy Pfeffer: BlueWall, yes deleting prims still has problems
[11:38] Nebadon Izumi: ive done more with physical prims than probably anyone
[11:38] Snoopy Pfeffer: autodelete leaves phantom prims
[11:38] Nebadon Izumi: this is an aincent problem
[11:38] Nebadon Izumi: this didnt start 2-3 weeks ago
[11:38] Nebadon Izumi: i can tell you that
[11:38] OtakuMegane Desu: Physics has been screwy in one way or another since the major work began though.
[11:38] Nebadon Izumi: try 2-3 months
[11:38] Snoopy Pfeffer: yes that is VERY old lol
[11:38] Dahlia Trimble: I didnt know autodelete worked at all
[11:38] Justin Clark-Casey: I don't think it was present in 0.6.7
[11:38] Snoopy Pfeffer: a year or more lol
[11:38] Nebadon Izumi: im pretty sure it was
[11:38] Nebadon Izumi: physical prims always sucked
[11:39] Nebadon Izumi: they always have floated off
[11:39] Justin Clark-Casey: nebadon: you mean physical prims or the inability to autoland?
[11:39] Snoopy Pfeffer: it should Dahlia
[11:39] Nebadon Izumi: physical prims
[11:39] Nebadon Izumi: landing is new
[11:39] Snoopy Pfeffer: but the objects come back after restarts
[11:39] Snoopy Pfeffer: and stay as phantoms
[11:39] Nebadon Izumi: phsyicla prims is aincent prohblem
[11:39] UUID Speaker: Math Dude: 5931012c-6a7e-2b43-e341-de313dd2dafe
[11:39] Strawberry Fride: 0.6.6 ninja physical prims float and move a bit - constant small adjustments - but regular physical prims, aside from being a bit of a pita, don't do that
[11:39] UUID Speaker: Math Dude: 5931012c-6a7e-2b43-e341-de313dd2dafe
[11:39] Justin Clark-Casey: yeah, I'm not sure about physical prims
[11:39] Nebadon Izumi: ive been doing 1000 prim physical stacks for months/years
[11:39] Nebadon Izumi: this is not new
[11:39] Dahlia Trimble: oh I dont know if physical prims are updated in region storage all that often
[11:39] Teravus Ousley: Yeah, I don't remember them floating off
[11:39] Nebadon Izumi: i do
[11:40] Nebadon Izumi: i do stacks almost weekly
[11:40] Nebadon Izumi: this is not a new issue
[11:40] Teravus Ousley: Previously, I could drag them and they might have had moments where they floated off sorta.. but they always came to rest
[11:40] Nebadon Izumi: it depends on the region load
[11:40] WhiteStar Magic: Nebadon actually buried me under a pile once... it was hillarious
[11:40] Nebadon Izumi: if i do 10 prims and its just me
[11:40] Nebadon Izumi: most dont
[11:40] Nebadon Izumi: if i do 500 prims
[11:40] Teravus Ousley: Unless they hit a region border.. or went under the terrain
[11:40] Nebadon Izumi: id say 25% float off
[11:40] Nebadon Izumi: or more
[11:40] Nebadon Izumi: for months
[11:41] Dahlia Trimble: last I tried there was still a problem with telling the viewer that a physical prim had stopped moving, several months ago
[11:41] Nebadon Izumi: it may have gotten worst lately
[11:41] Teravus Ousley: There are other side effects.. like I've outlined in my blog.
[11:41] Nebadon Izumi: but its not new at all
[11:41] Andress Renault: Is this the same as creating and object and it disappears? Does not land on the grid?
[11:41] Nebadon Izumi: try doing a 500 cube stack
[11:41] Teravus Ousley: notice how I go right through the physical cube
[11:41] Nebadon Izumi: on any revsion going back months
[11:41] Snoopy Pfeffer: lol
[11:41] Teravus Ousley: see?
[11:41] Justin Clark-Casey: that's so wrong :)
[11:41] Nebadon Izumi: ya
[11:41] BlueWall Slade: yes
[11:41] Strawberry Fride: floating physicals I have done same as you neb - 500-1000 prims. the floating stops if you relog though
[11:41] Andress Renault: Oooo..
[11:41] Justin Clark-Casey: you're a ghost!
[11:42] Snoopy Pfeffer: somewhere is the physical cube :)
[11:42] Snoopy Pfeffer: as phantom
[11:42] Nebadon Izumi: ya
[11:42] Nebadon Izumi: something is wrong with updates or something
[11:42] Nebadon Izumi: i have doors
[11:42] Nebadon Izumi: the doors downstairs on this building infact
[11:42] Nebadon Izumi: they open on collision
[11:42] Teravus Ousley: no, the cube is still physical bouncing around in physics.. but it's not bubbling up to the object update to the client.
[11:42] Nebadon Izumi: when you collide with them
[11:42] Nebadon Izumi: one time it works
[11:42] Nebadon Izumi: then after that the doors never work again
[11:42] Nebadon Izumi: i can tell you 100%
[11:42] Nebadon Izumi: that started before Jhurlimans work
[11:43] Andress Renault nods to Neb having noticed that.
[11:43] Snoopy Pfeffer: somewhere you can collide with an ghost for sure
[11:43] Nebadon Izumi: after the 1st collision
[11:43] Nebadon Izumi: the doors go phantom
[11:43] Nebadon Izumi: and never work again
[11:43] Nebadon Izumi: until sim is restarted
[11:43] Snoopy Pfeffer: with the new version things got worst
[11:43] Snoopy Pfeffer: worse*
[11:43] Nebadon Izumi: ya its possible things got worse
[11:43] Nebadon Izumi: but it didnt start recently
[11:43] Snoopy Pfeffer: door sometimes do not close visually
[11:43] Snoopy Pfeffer: but for the physics engine
[11:44] Andress Renault: Im seeing that box drifting into the ground.
[11:44] Nebadon Izumi: you can trigger it much faster by adding load to the region
[11:44] OtakuMegane Desu: Well, what else exactly are symptoms of this bug we're after?
[11:44] Snoopy Pfeffer: and once I did sink deep in the ground on a steep hill!
[11:44] Nebadon Izumi: if its an empty sim with 1 avatar
[11:44] WhiteStar Magic: same here with Hippo
[11:44] Nebadon Izumi: the chances of you triggering it are almost zero
[11:44] Nebadon Izumi: with less than 10 prims
[11:44] Nebadon Izumi: if you come here to Wright Plaza with 25 users
[11:44] Nebadon Izumi: its instant
[11:44] Nebadon Izumi: even with 1 prim
[11:45] Snoopy Pfeffer: once I was moving like this down in the ground
[11:45] Teravus Ousley: So that's one of the issue that I'm seeing anyway with recent OpenSim.
[11:45] Snoopy Pfeffer: so it is the same for objects and avatars
[11:45] Andress Renault: I disappear in my terrain once in awhile, just sink.
[11:45] Snoopy Pfeffer: yes
[11:45] Snoopy Pfeffer: I see that with the new version
[11:45] Teravus Ousley: Once an object gets in that state, it gets lost and no longer generates terse updates.
[11:45] Justin Clark-Casey: hum
[11:45] Teravus Ousley: That could lead to the rotation issues that you're describing.
[11:45] Snoopy Pfeffer: llSetPos has the same problems
[11:46] OtakuMegane Desu: Oh yeah
[11:46] Snoopy Pfeffer: not just rotations
[11:46] Hiro Protagonist: yep
[11:46] WhiteStar Magic: I reverted back to git# cedd827 - r11505 where I don;t sink into teh ground
[11:46] OtakuMegane Desu: Aligning my stuff was a damn pain lately
[11:46] Nebadon Izumi: also lldie
[11:46] Snoopy Pfeffer: even for non physical objects!
[11:46] Nebadon Izumi: produces wierd results
[11:46] Hiro Protagonist: I tried going back through IRC logs btw and find the convo I had with Melanie_T
[11:46] Nebadon Izumi: the prim is still visible
[11:46] Nebadon Izumi: but 100% unable to interact
[11:46] Nebadon Izumi: you can not move it
[11:46] Hiro Protagonist: but it had scrolled off the back of the buffer
[11:46] Nebadon Izumi: and if you do lldie
[11:46] Nebadon Izumi: while the prims are moving still
[11:46] Teravus Ousley: it's still bouncing around physics :)
[11:46] Nebadon Izumi: its 100 times worse
[11:47] Teravus Ousley: just lost
[11:47] Nebadon Izumi: i had a vehicle i was testing to poof when you stood
[11:47] Teravus Ousley: Hence the title of my blog entry
[11:47] Nebadon Izumi: if you stood while moving
[11:47] Teravus Ousley: "The Loss of references"
[11:47] Nebadon Izumi: it was absolute disaster
[11:47] Warin Cascabel: The interesting thing is, for the doors/elevators/etc. that don't update, if you put debugging statements in the scripts, they believe they've closed, moved, etc. - so I think it may have to do with updates going to the client, rather than not getting updated inworld.
[11:47] Nebadon Izumi: yes
[11:47] Nebadon Izumi: i agree
[11:47] Snoopy Pfeffer: yes Warin
[11:47] Nebadon Izumi: its something to client
[11:47] Nebadon Izumi: not simulator
[11:47] Nebadon Izumi: most of the time if you relog
[11:48] Nebadon Izumi: things have changed
[11:48] Teravus Ousley: the object update path is pretty complicated though
[11:48] Snoopy Pfeffer: yep
[11:48] Snoopy Pfeffer: I also experience that
[11:48] Adelle Fitzgerald: ive noticed when using llSetPos in a loop the prim will stop, but if you select and unselect it, the prim will move to the point where it should be when i selected it
[11:48] Snoopy Pfeffer: yes Adelle
[11:48] Snoopy Pfeffer: I did see the same
[11:48] Warin Cascabel: Yep. And usually - though not all the time - when it reaches the end of the loop, it'll update to the final position.
[11:48] Snoopy Pfeffer: it makes movements visible if you edit the object
[11:48] Teravus Ousley: objects have schedules and tests to pass before the update gets sent to the client.
[11:49] Snoopy Pfeffer: then they jump to the current position
[11:49] Snoopy Pfeffer: because of updates done
[11:49] Nebadon Izumi: one thing ive noticed too lately
[11:49] Nebadon Izumi: is if your cammed sorta far out
[11:49] OtakuMegane Desu: I have a chair with a follow script and the chair itself moves but the viewer doesn't update the visual position until you click on it
[11:49] Nebadon Izumi: you can see people logging in
[11:49] Nebadon Izumi: but if you cam to them
[11:49] Nebadon Izumi: they jump far away
[11:49] Nebadon Izumi: like they moved
[11:49] Nebadon Izumi: but your viewer didnt get the updates
[11:49] Nebadon Izumi: until you cam
[11:49] Nebadon Izumi: and not real far either
[11:49] Nebadon Izumi: im talking 20-30m
[11:50] Snoopy Pfeffer: yes Otaku, that is all that llSetPos / llSetRot bug
[11:50] Hiro Protagonist: I've not really had any probs like that
[11:50] Hiro Protagonist: the camera thing that is
[11:50] Nebadon Izumi: ya ive only noticed it a few times
[11:50] Hiro Protagonist: the positional updates and rotation updates yes
[11:50] OtakuMegane Desu: Cam I haven't had much issue with either
[11:50] Nebadon Izumi: but it seems like the viewer isnt getting al updates
[11:50] Nebadon Izumi: the times i did notice it too
[11:50] Hiro Protagonist: but only on non-physical objects
[11:50] Nebadon Izumi: the avatars were very primmy
[11:50] Nebadon Izumi: cause you see all the prims jump
[11:50] Nebadon Izumi: almost slide to the new position
[11:50] OtakuMegane Desu: The rest I've noticed plenty of :/
[11:50] Nebadon Izumi: in sequence
[11:51] Hiro Protagonist: I have seen physical objects sinking and/or falling through other prims though
[11:51] Lestat Furman is Online
[11:51] Dahlia Trimble is Offline
[11:51] Dorothea Lundquist is Online
[11:51] Nebadon Izumi: ya
[11:51] Nebadon Izumi: kittos car does the sink routing
[11:51] Nebadon Izumi: when you 1st sit on the car
[11:51] Nebadon Izumi: it sinks down a goof 4-5m
[11:51] Nebadon Izumi: then snaps back and is fine
[11:52] Nebadon Izumi: it takes a good 5-10 seconds
[11:52] Nebadon Izumi: then the car is fine after that
[11:52] Nebadon Izumi: but i find if you stand for driving for a while
[11:52] Nebadon Izumi: it crashes the sim
[11:52] Nebadon Izumi: almost 100% of the time
[11:52] Snoopy Pfeffer: wow
[11:52] OtakuMegane Desu: So basically a lot of thingsare not being updated unless something else interacts to force the update like a click.
[11:52] Snoopy Pfeffer: I had sim crashes because of tha llSetPos bug, too
[11:52] Andress Renault hums
[11:53] Snoopy Pfeffer: but not with the new version
[11:53] Snoopy Pfeffer: just with the old one
[11:53] Nebadon Izumi: at 1st i thought it was the mega region
[11:53] Teravus Ousley: well, as I said earlier, don't get discouraged. The issues are not unfixable. But they will take a concerted effort to resolve :)
[11:53] Nebadon Izumi: but then i tested the car in my sandbox
[11:53] Nebadon Izumi: and 1st try
[11:53] Nebadon Izumi: i crashed sim
[11:54] Hiro Protagonist: Teravus, anything I can do to help test, let me know
[11:54] Nebadon Izumi: oh ya im not discouraged no doubt we'll get past it
[11:54] BlueWall Slade: does anyone have a link they could share about digging through mono thread stacks ?
[11:54] BlueWall Slade: how to find deadlicks
[11:54] BlueWall Slade: locks, lol
[11:54] Justin Clark-Casey: mmm, deadlicks
[11:54] Nebadon Izumi: hehe
[11:54] BlueWall Slade: o.O
[11:54] Snoopy Pfeffer: hehe
[11:54] Andress Renault chuckles..
[11:54] Nebadon Izumi: how many deadlicks does it take to get to the center?
[11:54] OtakuMegane Desu: At least we have fairly stable regions to test on now lol
[11:54] Teravus Ousley: I know that the mono team released a commercial tool to debug mono applications from visual studio.
[11:55] OtakuMegane Desu: Could be a lot worse
[11:55] Adelle Fitzgerald: nebadon, am I ok to throw up a test region on the windows box to try and divide and conquer to get the commit/merge when this started?
[11:55] Justin Clark-Casey: nebadon: 5?
[11:55] Andress Renault: Well if your into that count it *laughs*
[11:55] Nebadon Izumi: lol ok Adelle
[11:55] Nebadon Izumi: ya that would be great
[11:55] Teravus Ousley: That's about the only way that I know, without compiling something into the assembly itself, to detect deadlocks.
[11:55] Adelle Fitzgerald: kk, ill jump on that shortly
[11:55] Hiro Protagonist: lmao
[11:55] Justin Clark-Casey: I just wait until it looks deadlocked and then painstakingly pour through the mono thread dump
[11:55] Justin Clark-Casey: that's why I'm so slooooowww :)
[11:55] Nebadon Izumi: ya we can do it no doubt
[11:56] Justin Clark-Casey: pore not pour
[11:56] Adelle Fitzgerald: im not sure if i will find it, but i will certainly try
[11:56] Nebadon Izumi: it just might take a bit
[11:56] Snoopy Pfeffer: did anybody also experience that bug in os search causing viewer crashes?
[11:56] BlueWall Slade: will a dead heartbeat show up as a deadlock ?
[11:56] Adelle Fitzgerald: ive seen that, yes
[11:56] Snoopy Pfeffer: for example search "rent" in the search land window here!
[11:56] Snoopy Pfeffer: hehe
[11:56] BlueWall Slade: @Snoopy yes
[11:57] Teravus Ousley: viscous pour? or viciously pour?
[11:57] Snoopy Pfeffer: but that will crash your viewer
[11:57] Adelle Fitzgerald: just search for 'free' in OSgrid and its sure to crash the viewer
[11:57] Justin Clark-Casey: bluewall: afaik, that's what I would expect
[11:57] Andress Renault: Yes Ive seen it go blank..map goes blank
[11:57] Snoopy Pfeffer: yes
[11:57] Snoopy Pfeffer: if there are too many matches search crashes the viewer
[11:57] Justin Clark-Casey: heh - crash the freeloaders.... :)
[11:57] Snoopy Pfeffer: "free", "rent" and probably many search strings cause that
[11:57] Andress Renault is definately not in that category.
[11:58] Snoopy Pfeffer: did you all crash? hehe
[11:58] WhiteStar Magic: curious, what viewers are you guys using when doing those searches ?
[11:58] Hiro Protagonist: nope
[11:58] Andress Renault: Hippo here
[11:58] Hiro Protagonist: I didnt search tho hahaha
[11:58] Adelle Fitzgerald: i didnt risk it ;)
[11:59] Nebadon Izumi: ive seen it too
[11:59] Hiro Protagonist: I'm running emerald atm
[11:59] Nebadon Izumi: i probably need to purge the search DB
[11:59] Nebadon Izumi: let it re-index
[11:59] Snoopy Pfeffer: I did try it with all kinds of viewers
[11:59] Andress Renault: Any other server I get graphics problems on a Alien 51 box.
[11:59] Snoopy Pfeffer: it is always crashing for me
[11:59] Snoopy Pfeffer: under windows and linux
[11:59] Snoopy Pfeffer: with Hippo, SL, Emerald, merkat
[11:59] Nebadon Izumi: ya its probably wierd character or something maybe in the DB
[11:59] Nebadon Izumi: the search module could stand updating
[12:00] Nebadon Izumi: it seems to not work a lot
[12:00] Justin Clark-Casey: alright, I'm off - I'm glad we got some discussion on some of the issues even if we're a little way from resolving them.
[12:00] Justin Clark-Casey: buy folks
[12:00] BlueWall Slade: free weird characters???
[12:00] Justin Clark-Casey: er, bye
[12:00] WhiteStar Magic: That happened a couple of months back too neb and you wound up having to purge teh DB cause someone started poutting in * and / etc
[12:00] Nebadon Izumi: later Justincc
[12:00] BlueWall Slade: buy??
[12:00] Dorothea Lundquist is Offline
[12:00] Snoopy Pfeffer: bye Justin
[12:00] Nebadon Izumi: hehe SOLD! to the man in the red shirt
[12:00] Andress Renault: Buy Justin how much! :)
[12:00] WhiteStar Magic: See ya JCC
[12:00] Nebadon Izumi: you can pickup your crowd at the main desk
[12:00] Andress Renault: Oh darn missed
[12:00] Justin Clark-Casey: my brain is scrambled :)
[12:01] Nebadon Izumi: hehe
[12:01] BlueWall Slade: now where is that opencollar
[12:01] OtakuMegane Desu: Ok, Cool SL gets crashed by search too lol
[12:01] Andress Renault: Saw one on the marketplace
[12:01] Justin Clark-Casey is Offline
[12:01] Snoopy Pfeffer: btw if anybody is interested in trying the newest version of the paypal module: http://github.com/SnoopyPfeffer/Mod-PayPal
[12:01] Eryn Galen is Offline
[12:01] Snoopy Pfeffer: it is a fork of Adam's version
[12:01] Andress Renault: Thanks for that Snoopy
[12:02] Snoopy Pfeffer: we did discuss that it is the best if I go forward with a forked repository
[12:02] Snoopy Pfeffer: welcome
[12:02] Penny Lane: Cheers for link, Snoopy
[12:02] BlueWall Slade: nice meeting. Thanks everybody
[12:02] BlueWall Slade is Offline
[12:02] Snoopy Pfeffer: the wiki contains a Current Status page
[12:02] Snoopy Pfeffer: it lists all that has been done and still needs to be done
[12:03] Andress Renault: Right now Im just going to set up a webpage and put buttons on it, Ive worked with Paypal carts before.
[12:03] Teravus Ousley: ok, I need to go also. Thanks for being at the meeting.
[12:03] Penny Lane: Nice and stable, it's been a pleasure here today. And the beta-2 of Imprudence has stayed up 100% of the hour, neat.
[12:03] Snoopy Pfeffer: me too :)
[12:03] Hiro Protagonist: Thanks for talking with us today Teravus
[12:03] Snoopy Pfeffer: thanks and see you all soon again :)
[12:03] Andress Renault: Thanks all.
[12:03] Nebadon Izumi: can someone send me the log?
[12:03] Nebadon Izumi: i missed 20 minutes
[12:04] Nebadon Izumi: just drop a notecard on me if you do
[12:04] Strawberry Fride: thanks all - seeya later :)
[12:04] Penny Lane: I'll try, Neb

Personal tools
General
About This Wiki