Chat log from the meeting on 2013-06-11

From OpenSimulator

Jump to: navigation, search
[11:01] Justin Clark-Casey: hello folks
[11:02] Nebadon Izumi: hey Justin
[11:02] logger sewell: Hi Justin
[11:02] Nebadon Izumi: must have been logging in right as I messaged you on skype :)
[11:02] Andrew Hellershanks: I crashed Singularity again by switching workspaces. :P
[11:02] Justin Clark-Casey: yes,had to route around the jump points
[11:03] Unknown UserUMMTGUN2: Hi Justin
[11:03] BlueWall.Slade Hello
[11:05] Justin Clark-Casey: anything opensim to discuss today?
[11:05] Nebadon Izumi: I have something odd thats happening to me
[11:05] Nebadon Izumi: let me show you
[11:05] Andrew Hellershanks: I was having an issue with groups but I moved the webserver and the problem seems to have cleared up so I think it was a network issue.
[11:05] Nebadon Izumi: see the cube attached to my hand
[11:05] Justin Clark-Casey: yes
[11:05] Nebadon Izumi: watch what happens when i detach it
[11:06] Cuteulala Artis is online.
[11:06] Nebadon Izumi: shit
[11:06] Nebadon Izumi: didnt do it
[11:06] Nebadon Izumi: hrmm
[11:06] Andrew Hellershanks: :)
[11:06] Nebadon Izumi: guess its random
[11:07] Justin Clark-Casey: my favourite kind of bug
[11:07] Nebadon Izumi: sim freeze?
[11:07] Vivian Klees: if it detached it should go in front of you
[11:07] Nebadon Izumi: and it says (Attached to Invalid Point) in my inventory
[11:07] Justin Clark-Casey: yes
[11:07] Richardus Raymaker: lag
[11:07] Andrew Hellershanks: Detaching something it should go away. Dropping it will keep it in the region
[11:07] Vivian Klees: dropped I mean
[11:07] Nebadon Izumi: grr
[11:07] Justin Clark-Casey: it says that now?
[11:07] Richardus Raymaker: hmm no where's my post
[11:07] Nebadon Izumi: anyway ive been seeing it a lot when i detach something from my hand
[11:07] Nebadon Izumi: everything on my head comes off
[11:08] Nebadon Izumi: but in inventory looks like its still attached
[11:08] Nebadon Izumi: but it says (Invalid Attachmnet Point)
[11:08] Nebadon Izumi: where it should say what point its attached to
[11:08] Nebadon Izumi: let me try something more complex than a cube
[11:08] Justin Clark-Casey: since when have you been seeing this?
[11:08] Richardus Raymaker: i see in my logs this "OpenSim.Region.CoreModules.Framework.UserManagement.UserManagementModule [SCENE]: Unable to parse Uri http:\\www." and someone else with 0.7.6dev complained about it to. thats not good if you cannot retrieve the profile info from the source grid. notsure if there are other side effects.
[11:08] Fred.Appleby I have seen that also neb
[11:08] Nebadon Izumi: couple weeks
[11:08] Cuteulala Artis is offline.
[11:09] Nebadon Izumi: hmm
[11:09] Nebadon Izumi: wierd its not doing it now
[11:09] Andrew Hellershanks: gbbiab
[11:09] Cuteulala Artis is online.
[11:09] Justin Clark-Casey: you have a low bar for wierd
[11:09] Nebadon Izumi: heh
[11:10] Nebadon Izumi: maybe its something to do with hypergrid travels
[11:10] Nebadon Izumi: let me HG out and come right back
[11:10] Nebadon Izumi: 1 sec
[11:10] Justin Clark-Casey: possibly, or even the inventory changes
[11:10] Justin Clark-Casey: though it sounds like it was occuring before that
[11:10] Richardus Raymaker: the error i get happens when you used the --profile tag with oar
[11:10] Justin Clark-Casey: richardus: I'm not surprised if that is buggy
[11:10] Nebadon Izumi is online.
[11:11] Simulator Version v0.5 shouts: OpenSim 0.7.6 Dev          3a722ef: 2013-06-07 23:59:02 +0100 (Unix/Mono)
[11:11] Nebadon Izumi: AHA
[11:11] Vivian Klees: bye bye hair
[11:11] Nebadon Izumi: do you see everything missing from my head?
[11:11] Nebadon Izumi: its HG related
[11:11] Justin Clark-Casey: yes
[11:11] Richardus Raymaker: your hair
[11:11] Nebadon Izumi: if you HG out and back in
[11:11] Nebadon Izumi: then dettach something
[11:11] Teravus Ousley is online.
[11:11] Nebadon Izumi: it detaches the object and everything on your head
[11:11] Dahlia Trimble is online.
[11:11] Justin Clark-Casey: so, an issue for crista
[11:11] Nebadon Izumi: i didnt even do anything when i HG'd out
[11:11] Unknown UserUMMTGUN: Neb, you look grey and nekkid
[11:12] Richardus Raymaker: did not somneone complained about that last week to ?
[11:12] Nebadon Izumi: not sure who that is your name is Uknown
[11:12] Nebadon Izumi: whoever said im just grey
[11:12] Unknown UserUMMTGUN: Was me, Pathfinder
[11:12] Richardus Raymaker: Pathfinder.Lester Neb, you look grey and nekkid
[11:12] Teravus Ousley: sorry about the fashionably lateness :)
[11:12] Richardus Raymaker: hi teravus
[11:12] Justin Clark-Casey: hello teravus
[11:12] Teravus Ousley: hallo
[11:12] Deanna Janus: hi all
[11:12] Justin Clark-Casey: hi deanna
[11:12] Fred.Appleby did you have boots neb?
[11:13] Fred.Appleby oh, now i see boots
[11:13] Richardus Raymaker: well i see shoes on neb
[11:13] Nebadon Izumi: its wierd its not everything on your head it seems
[11:13] Nebadon Izumi: its just the Skull Attachment point
[11:13] Andrew Hellershanks: neb looks fine to me except his head seems a bit small for the body
[11:13] Nebadon Izumi: as soon as i reatttach the skull part
[11:13] Nebadon Izumi: everything on my head comes back
[11:13] Nebadon Izumi: my hair is attached to "Skull" point
[11:13] Nebadon Izumi: odd one
[11:14] Nebadon Izumi: i'll let diva know
[11:14] Nebadon Izumi: well as soon as shes back online I will :)
[11:15] Richardus Raymaker: maby someone can figure a way out so you can always retrieve profile info from grid to. if HG is disabled / bocked or so. or maby other way
[11:15] Nebadon Izumi: thats only complaint I have this week :)
[11:15] Justin Clark-Casey: really, just one?
[11:15] Nebadon Izumi: haha
[11:16] Justin Clark-Casey: those are the worst - I can never remember why the veiewer ends up with that 'invalid attachment point' stuff
[11:16] Justin Clark-Casey: the wrong message gets sent out and confuses it
[11:16] Richardus Raymaker: hmm. thats long time i heared that one
[11:16] Nebadon Izumi: ya maybe something with serializing the layers after an HG
[11:16] Nebadon Izumi: for multiple attach points?
[11:16] Nebadon Izumi: wierd that hand would effect skull
[11:16] Nebadon Izumi: almost like somethings out of order
[11:16] Justin Clark-Casey: is it just skull
[11:17] Nebadon Izumi: ya
[11:17] Justin Clark-Casey: skull is attach point zero, which is a special number
[11:17] Nebadon Izumi: but everything on my head comes off visibly
[11:17] Unknown UserUMMTGUN2: FS and Sing haave incompatible attach points
[11:17] Nebadon Izumi: but as soon as i reattach skull
[11:17] Nebadon Izumi: everything on head is visible again
[11:17] Richardus Raymaker: wuhmm, so that sign is still right. "yalk to the hand , the head dont listen" :D
[11:17] Nebadon Izumi: and skull is the only one that says Invalid attach point
[11:19] Justin Clark-Casey: I love attachment issues
[11:19] Nebadon Izumi: heh ya
[11:20] Nebadon Izumi: anything that touches inventory sucks
[11:20] Vivian Klees: anything on terrain editing?
[11:20] BlueWall.Slade nope
[11:21] Justin Clark-Casey: that's a 'nice' simple answer :)
[11:21] Nebadon Izumi: haha
[11:21] BlueWall.Slade lol
[11:21] Nebadon Izumi: someone needs to enable a throttle
[11:21] Nebadon Izumi: basically
[11:21] Nebadon Izumi: same as inventory
[11:21] Nebadon Izumi: viewer is throwing too many packets at simulator
[11:21] BlueWall.Slade will have to experiment with it
[11:21] Nebadon Izumi: and simulator is accepting all of them
[11:21] BlueWall.Slade need to drop requests
[11:21] Nebadon Izumi: it should be discarding like 90% of them
[11:21] Teravus Ousley: Just a note.. if we can identify in the client stack that you're the one editing the terrain, we can loosen the terrain throttle for the editing avatar
[11:21] Richardus Raymaker: does terrain edit strength work now ?
[11:21] Teravus Ousley: and that would solve it
[11:21] Fred.Appleby Each time attach from my suitcase, a new link in my suitcase root is created
[11:22] Andrew Hellershanks: why is viewer sending so many packets if it doesn't really need all of them?
[11:22] Nebadon Izumi: that would be a question for the brilliant engineers over at SL
[11:22] Nebadon Izumi: lol
[11:22] Teravus Ousley: The viewer sends lots of packets that we don't need :)
[11:22] Unknown UserUMMTGUN2: ahh so thats why some viewers are better at terrain editing then others
[11:22] Justin Clark-Casey: I think nebadon means server is throwing packets at viewer, perhaps
[11:22] Andrew Hellershanks: :)
[11:22] Nebadon Izumi: no other way around
[11:23] Richardus Raymaker: Nebadon, would decrease viewer bandwidth to lowest help with terrain edit ?
[11:23] Nebadon Izumi: when you click the paint brush to edit terrain
[11:23] Nebadon Izumi: it sends out a massive amount of packets to simulator
[11:23] Richardus Raymaker: aha
[11:23] Teravus Ousley: Yes, I do enjoy a bit of sarcasm
[11:23] Justin Clark-Casey: oh ok
[11:23] BlueWall.Slade if you "paint" the tools over the surface, it generates a massive amount of requests
[11:23] Justin Clark-Casey: so one needs an adaptive algorithm
[11:23] Nebadon Izumi: and we are accepting all of them
[11:23] Nebadon Izumi: Melanie has a good understanding as does Teravus it sounds like :)
[11:23] Dahlia Trimble is offline.
[11:23] BlueWall.Slade once you do that fgor a few sec. then it takes a while to watch the editing finish
[11:24] Teravus Ousley: well, it could be solved two ways
[11:24] Richardus Raymaker: yes
[11:24] Teravus Ousley: send less terrain patches
[11:24] Teravus Ousley: or
[11:24] Teravus Ousley: open up the terrain patch throttle for the editing user
[11:24] Richardus Raymaker: well, if the send less packets is thge easy way to fix it for now it would be improvement i think ?
[11:24] BlueWall.Slade Melanie says that in SL the distance is factored into it too
[11:25] dan banner: i think editing terrain is painfully slow already
[11:25] Teravus Ousley: maybe Richardus.
[11:25] BlueWall.Slade RiRa, it's accepting less packets to process
[11:25] Richardus Raymaker: hi dan, well it worked pretty fast and nice. for the first 3 seconds
[11:25] Justin Clark-Casey: teravus: conceivably one could send self packets via a different udp queue than the terrain one
[11:25] Richardus Raymaker: maby 10seconds
[11:25] Teravus Ousley: That is true.
[11:26] Richardus Raymaker: after that edit always run behind
[11:26] Justin Clark-Casey: and then pass in the UUID of the agent actually making the changes to decide which queue it should be in
[11:26] dan banner: hi rich
[11:26] Teravus Ousley: It's just that editing the terrain jams them into the UDP throttle
[11:26] BlueWall.Slade if you have ever seen a video of cattle being loaded onto a train from a stockyard...
[11:26] Teravus Ousley: so you get 2 seconds before it fills up :)
[11:26] Teravus Ousley: then it starts delaying
[11:26] Justin Clark-Casey: interesting - so a "show queues" should reveal that
[11:26] Teravus Ousley: Yes
[11:26] BlueWall.Slade possibly
[11:26] Unknown UserUMMTGUN2: need a bigger boxcar then ;)
[11:26] Teravus Ousley: That's the cause of that delay issue
[11:27] Teravus Ousley: so either loosen up the throttle on the editing user
[11:27] BlueWall.Slade no Arielle - eat less meat :p
[11:27] Teravus Ousley: or send less terrain patches
[11:27] Vivian Klees: a lot of sand for the hourglass
[11:27] Unknown UserUMMTGUN2: a girl cant live on veggies alone
[11:28] Justin Clark-Casey: teravus: now I think about it, the last unknown user change I made aimed to always add the user binding as early as possible
[11:28] Justin Clark-Casey: in fact, when the avatar first contacted the simulator before any attachmetns could be sent
[11:28] Justin Clark-Casey: but that has not appeared to help
[11:28] Justin Clark-Casey: commit cc7aa88
[11:28] Justin Clark-Casey: on may 29
[11:29] Panther Lionheart: sorry everyone. didn't meant to land on you
[11:29] Richardus Raymaker: hi robert. have question in while for you
[11:29] Teravus Ousley: what if the viewer is asking a different region for your name<--->uuid mapping
[11:29] dan banner: i havent hardly seen any unknown users
[11:29] Nebadon Izumi: still not sure who it is
[11:29] Unknown UserUMMTGUN2 looks to see which unknown user said that
[11:29] tx Oh: do we have a load test later?
[11:29] Robert Adams: when editing terrain it does indeed sent out way more updates than necessary.... I looked at the code but it wasn't a quick fix
[11:29] Justin Clark-Casey: teravus: unless it's a region that the viewer has never been on that shouldn't be a problem
[11:29] Nebadon Izumi: I see one here today
[11:30] Richardus Raymaker: Robert, does bullet already have llSetKeyframedMotion ?
[11:30] Robert Adams: what's up Richardus
[11:30] Teravus Ousley: The viewer will only ask for it once and then cache it for a duration of time.
[11:30] Unknown UserUMMTGUN2: richard and Dan are unknown to me
[11:30] Justin Clark-Casey: teravus: I thought the cache was permanent?
[11:30] Richardus Raymaker: noo, i know you 4 years ! :)
[11:30] Justin Clark-Casey: teravus: not the case?
[11:30] Teravus Ousley: Just a note, KeyframedMotion as implemented in Melanie's commit is non-physical movement
[11:30] Unknown UserUMMTGUN2: and Vivian
[11:30] Robert Adams: key frame motion is implemented totally in the simulator and will work with any physics engine
[11:30] Teravus Ousley: I think it's effectively forever Justin
[11:30] Richardus Raymaker: well is it smooth teravus ?
[11:31] Teravus Ousley: It's 'smooth-ish'
[11:31] Justin Clark-Casey: teravus: which makes this proiblem even harder to detect :0
[11:31] Justin Clark-Casey: I mean, detect a solution
[11:31] dan banner: ari are you on the latest code ?
[11:31] Richardus Raymaker: ok. thats cool. i have plans to test 0.7.6dev on standalone.
[11:31] Robert Adams: it looks like its clock is 20 times per second which is faster than the physics engine... although I haven't tested it
[11:31] Justin Clark-Casey: but it started happening at some point in time - we haven't always had these unknown user issues
[11:31] Unknown UserUMMTGUN2: from my standalone? No. Viewer is pretty up to date though
[11:31] Teravus Ousley: That's true..
[11:32] Teravus Ousley: We should make it a point to tell people to clear their cache before coming to a load event
[11:32] Justin Clark-Casey: that would cause... a lot more load :)
[11:32] Nebadon Izumi: ya
[11:32] Unknown UserUMMTGUN2: heh yes
[11:32] Teravus Ousley: Yes, it would
[11:32] Richardus Raymaker: lol
[11:32] Robert Adams: if I stand behind the bench, justin is just out of chat range
[11:32] Nebadon Izumi: its unrealistic
[11:32] Unknown UserUMMTGUN2: and a lot o clouds
[11:32] Nebadon Izumi: to have everyone clear their cache
[11:32] vegaslon plutonian: only problem I see with llSetKeyFramedMotion on bulletsim is it moves stuff so fast the physics engine does not react fast enought to it so you end up dropping through anything you have moving
[11:32] Nebadon Izumi: even if i asked everyone to
[11:32] Nebadon Izumi: not everyone will
[11:33] Teravus Ousley: well. Not really. How many people are going to have been to the conference region prior?
[11:33] Nebadon Izumi: everyone who came last week
[11:33] Nebadon Izumi: haha
[11:33] Richardus Raymaker: sound slike its useing the same system as at_target. that can be way to fast to
[11:33] Unknown UserUMMTGUN2: but i suppose it would put a good load on the server
[11:33] Richardus Raymaker: well. i need to find time to play with it robert.
[11:34] Teravus Ousley: I'm just saying.. if it can handle cache-free users.. it should stand up well to cache full users.
[11:34] Robert Adams: I think the keyframe code just changes position... could go right through things
[11:34] Richardus Raymaker: hmm. it would be problem if its sinking in prims.
[11:34] Richardus Raymaker: otherwise i try moveto again
[11:34] Teravus Ousley: Yes, that's one of the issues with the way keyframe is implemented now.
[11:34] Richardus Raymaker: in the hope thats working better now
[11:34] Robert Adams: that's also a known problem (and a TODO) with llMoveToTarget and BulletSim (only sets position when it should be setting forces)
[11:34] Teravus Ousley: KeyframeMotion on SL relies on a Havok Keyframe function
[11:35] Teravus Ousley: So.. until someone implements it in the physics layer on a different physics engine..
[11:35] BlueWall.Slade yeah I noticed that - the LL docs say it's good for elevators, but the avatar falls throuhg
[11:35] BlueWall.Slade it happens in ODE too
[11:35] Richardus Raymaker: not played enough with that command there. only small test
[11:35] Teravus Ousley: It's basically llSetPos in tiny increments
[11:35] Ai Austin is online.
[11:35] tx Oh: this keyframe thing, is it a new feature? i think i've never heared it befor ?-)
[11:35] Teravus Ousley: or an abstraction layer to make a large distance llSetPos look smooth
[11:35] Robert Adams: yep, Teravus
[11:35] Richardus Raymaker: yes.. but it dont load the lsl engine
[11:36] Richardus Raymaker: with llsetpos and at_taget you can push your cpu pretty high
[11:36] Richardus Raymaker: uhmm no parametersfast not llsetpos
[11:36] Robert Adams: and not a general solution... events should have been added in core and keyframing should have been a region module
[11:38] Teravus Ousley: so my suggestion for keyframing.. is have a hook that asks the physics engine if it supports that kind of thing.. and if so.. pass of the responsibility to the physics engine. If not.. llSetPos in small increments like it is now.
[11:39] Justin Clark-Casey: that doesn't sound unreasonable to me
[11:39] Justin Clark-Casey: and if something can be a region module then it should
[11:39] Richardus Raymaker: llsetpos or parametersfast ? the last one can do it smoother :O
[11:39] Vivian Klees: how small are those increments out of curiosity
[11:39] Teravus Ousley: The actual call that it uses is neither :) It just moves the object
[11:39] Teravus Ousley: vivian, very small :)
[11:39] Richardus Raymaker: lsetpos havbe delay. thats why
[11:40] Teravus Ousley: small enough to look smooth
[11:40] Ai Austin is offline.
[11:40] Richardus Raymaker: i think steps of .1 is already smooth if you do it in lsl
[11:40] Robert Adams: I want to allow physical type interactions implementable outside the physics engine
[11:41] Robert Adams: one reason I added the StartOfHeartbeat event
[11:41] BlueWall.Slade if the physics engine did the moving, then it would be one continuous movement?
[11:41] vegaslon plutonian: and it would effect the physical objects in a way to move stuff with it
[11:41] Teravus Ousley: if the physics engine did, then it would interact physically with the physics world.. would be the main difference
[11:41] Robert Adams: no reason a region module can't compute and set a force on an object and then let the physics engine move it
[11:41] Unknown UserUMMTGUN2: crashy here today
[11:42] Richardus Raymaker: that sounds good robert. and for me logic
[11:42] Justin Clark-Casey: nebadon: seem to see a fair few errors in the wright plaza log about fetching inventory
[11:42] Justin Clark-Casey: all document empty errors by the look of it
[11:43] Nebadon Izumi: thats normal
[11:43] Nebadon Izumi: i dont see the errors Justin
[11:43] Nebadon Izumi: did they stop
[11:43] Nebadon Izumi: probably an HG visitor
[11:43] Justin Clark-Casey: "[XINVENTORY SERVICES CONNECTOR]: Exception in GetItem: System.Xml.XmlException: Document element did not appear. Line 1, position 1."
[11:44] Justin Clark-Casey: maybe it is hg
[11:44] Nebadon Izumi: ya unfortunately inventory errors/warnings are pretty prevalent
[11:44] Richardus Raymaker: your wheels are stolen.
[11:44] Nebadon Izumi: its probably a HG visitor
[11:44] Robert Adams: looks good on you, Teravus
[11:44] Nebadon Izumi: if you look 1 line above that Justin
[11:44] Nebadon Izumi: its a HG visitor
[11:44] Teravus Ousley: :)
[11:44] Nebadon Izumi: I think anyway
[11:45] Nebadon Izumi: request 113582 POST took 163127ms, 63127ms writing, ID=df5c59c4-db03-4848-829f-1381f0
ERVICES CONNECTOR]: Exception in GetItem: System.Xml.XmlException: Document element did not appear. Line 1, position 1.
ader.Read () [0x0017b] in /root/install/mono-3.0.7/mono-3.0.7/mcs/class/System.XML/System.Xml/XmlTextReader.cs:579 
eader.Read () [0x0006b] in /root/install/mono-3.0.7/mono-3.0.7/mcs/class/System.XML/System.Xml/XmlTextReader2.cs:566 
ent.ReadNodeCore (System.Xml.XmlReader reader) [0x00035] in /root/install/mono-3.0.7/mono-3.0.7/mcs/class/System.XML/System.Xml/XmlD
ent.ReadNode (System.Xml.XmlReader reader) [0x00032] in /root/install/mono-3.0.7/mono-3.0.7/mcs/class/System.XML/System.Xml/XmlDocum
ent.Load (System.Xml.XmlReader reader) [0x00019] in /root/install/mono-3.0.7/mono-3.0.7/mcs/class/System.XML/System.Xml/XmlDocument.
ent.LoadXml (System.String xml) [0x0002d] in /root/install/mono-3.0.7/mono-3.0.7/mcs/class/System.XML/System.Xml/Xm
[11:45] Nebadon Izumi: oops
[11:45] Nebadon Izumi: anyway
[11:45] Nebadon Izumi: the uuids match
[11:45] Nebadon Izumi: i didnt copy the whole thing
[11:45] Ai Austin is online.
[11:45] Nebadon Izumi: if you look at the big red block though and the line just above it
[11:45] Nebadon Izumi: the asset uuids match
[11:45] Ai Austin is offline.
[11:46] Nebadon Izumi: teravus is living in a van down by the river?
[11:46] Nebadon Izumi: smoking your doobies!
[11:46] Justin Clark-Casey: well, tehcnically those are separate log entries but yes, in this case they are ver likely connected
[11:46] Nebadon Izumi: the uuids match exactly
Personal tools
About This Wiki