Chat log from the meeting on 2016-12-13

From OpenSimulator

Jump to: navigation, search

[11:03] Andrew.Hellershanks I was making some minor updates to the OS wiki. I didn't remember it was Tuesday until my PDA reminded me.
[11:03] James.atLLOUD I feel like I've seen most everyone here just days ago!
[11:03] George Equus: :)
[11:04] Andrew.Hellershanks hehe, yea, there are a few people here that we might have seen in the last couple of days.
[11:04] Selby.Evans We met at OSCC, of course
[11:04] Andrew.Hellershanks I didn't make it to many panels this year.
[11:05] George Equus: OSCC... now, what was that again... hmm
[11:05] George Equus: Oh yes! remember
[11:06] James.atLLOUD :)
[11:06] Andrew.Hellershanks If there is a panel you wanted to attend but were unable to do so the panels will be available online soon. Videos of last years panels were made available on the website not long after the previous event.
[11:07] George Equus: Been looking for them, bit early yet I suppose
[11:07] James.atLLOUD Yes, speaking for the video team, it was an intense effort. Hope to get them uploaded soon.
[11:07] Andrew.Hellershanks yea. I don't know how soon after the conference they will be available. I'd give it a few more days.
[11:08] Ubit Umarov: you can cut my parts off on video edit :p
[11:08] Andrew.Hellershanks The team did a good job of doing smooth transitions between panel audio and switching to music between events.
[11:09] Kayaker.Magic Some of us, you can make the video half as long by cutting out all the times I said Umm...
[11:09] Andrew.Hellershanks Last year there were some audio related hiccups. I was not aware of any this year when I was at the OSCC
[11:09] Andrew.Hellershanks Hello, Robert. I thought that was you I saw flying in.
[11:09] James.atLLOUD Inworld ironically is a big challenge.
[11:09] Andrew.Hellershanks Kayaker, Could turn it in to a drinking game. :)
[11:10] James.atLLOUD inworld audio
[11:10] George Equus: Presentations felt very smooth this year, at least those I attended
[11:11] Andrew.Hellershanks I told a few people about the OSCC. At least one of them was watching the live conference stream via the website instead of being inworld.
[11:11] Andrew.Hellershanks George, indeed.
[11:13] Selby.Evans Eveything I saw went well --including Shakespeare
[11:13] Andrew.Hellershanks It went very well. I think it is all the more impressive when I think the staff didn't have as much time to prepare for the event as last year.
[11:14] Kayaker.Magic I saw a lot of the actors in the Shakespeare event revert to a sitting pose when they were not actively moving. Was that just my viewer?
[11:14] James.atLLOUD I am so glad there were no 'tracks' . same time presentations seem un-necessary online.
[11:14] Robert Adams: the conference did go very smoothly from both a technical and an organizational point of view
[11:15] Selby.Evans Kayaker -- the actors sat to reduce load on the server.
[11:15] Andrew.Hellershanks I was wanting to be on the Keynote panel with the other core developers at the start of the OSCC but RL activities prevented me from attending.
[11:16] Kayaker.Magic I'd see an actor walking across the stage, stop for a second, sit on thin air, walk correctly again, stop and sit on thin air. Other actors did not show this behavior.
[11:16] James.atLLOUD We missed you Andrew, but certainly understand.
[11:17] Andrew.Hellershanks I was at the VIP Q&A but I didn't have anything to say with the topics discussed.
[11:17] Ubit Umarov: old opensim issue possible
[11:17] Simulator Version v0.5 ruft: OpenSim Dev 7d8cb24: 2016-12-12 19:16:51 +0000 (Unix/Mono)
[11:18] Ubit Umarov: ( the unsit not working )
[11:18] Kayaker.Magic Does anyone know that physics engine is used here in Wright Plaza?
[11:18] Ubit Umarov: bullet
[11:19] Andrew.Hellershanks I just had a thought. Has anyone tried a sim crossing in a grid where one region is on BulletSim and the other is on ubODE?
[11:19] Misterblue Waves: I have... and it works!
[11:19] Andrew.Hellershanks I wonder if there would be any issues in making a crossing involving a change in physics.
[11:19] Misterblue Waves: most of the crossing logic is in core rather than in the physics engines themselves
[11:19] Andrew.Hellershanks Misterblue, cool. Have you tried it with vehicles?
[11:20] Andrew.Hellershanks Misterblue, true. Unless a vehicle is involved in the crossing the physics engine is unlikely to have much impact.
[11:20] Misterblue Waves: yes... one of Nebadon's cars is my tester
[11:21] Andrew.Hellershanks Hm... I wonder if Nebadons car uses physics.
[11:22] Andrew.Hellershanks I want to get a touring balloon working one day. I haven't had time to look in to what script changes are needed to make the balloon move properly.
[11:22] Andrew.Hellershanks At the moment it might as well be a lead balloon as it won't even get off the ground.
[11:23] Misterblue Waves: it does.... it has a cube/sphere physical center with phantom meshes for the body
[11:23] George Equus: Andrew, drop some ballast!
[11:23] Andrew.Hellershanks George, :)
[11:23] George Equus: I usually only bring my tooth brush..
[11:24] George Equus: :)
[11:24] Andrew.Hellershanks Misterblue, ah, yes. I remember that now. I haven't looked at his car in a while.
[11:24] Ubit Umarov: i use the one still with phycis for old ode
[11:24] Ubit Umarov: and 253 physical prims some scultps
[11:25] Ubit Umarov: nice physics stress thing
[11:25] Ubit Umarov: sit/unsit on it
[11:25] George Equus: sounds horrific..
[11:26] Ubit Umarov: OKC Racer v0.186X (heavy)
[11:27] Ubit Umarov: Robert did had nice physics explosions with it if a remember
[11:27] Ubit Umarov: ( well so did i oops :p )
[11:28] George Equus: It was nice to see some VIP ppl attend the closing party, Crista being one of them. cool!
[11:29] Andrew.Hellershanks I dropped in to the event at the Maritime Club for a short while but i didn't stay long.
[11:29] Robert Adams: I am sad I missed the closing party -- RL gets in the way of so many fun things
[11:30] Andrew.Hellershanks So true, Robert
[11:30] Ubit Umarov: ohh you missed the naked andy ?
[11:31] George Equus: Nazirah managed to get and stay online too. Poor thing, horrible connection...
[11:31] Andrew.Hellershanks I was just looking over some of the recent changes to OpenSim this past week.
[11:31] Andrew.Hellershanks One of the changes that I know about is some improvements to handling of avatar profiles for NPCs.
[11:32] Andrew.Hellershanks The changes were made in both core profiles and in the addon OpenSimProfile module.
[11:33] Andrew.Hellershanks Users can also set the AboutText for an NPC's profile using a new os function.
[11:34] Ubit Umarov: kayaker.Magic andrew told me you where going to retest guns aim on 0.91 ?
[11:34] Kayaker.Magic Yes, I was doing that this morning, not finished yet but first indications are that it is working good on 0.9 with ubODE.
[11:35] Kayaker.Magic I want to try with more combinations
[11:35] Ubit Umarov: well robert is a big target
[11:35] Andrew.Hellershanks Kayaker, It would also be good to know if it has been fixed in BulletSim
[11:36] Kayaker.Magic Can't test here, llCastRay V1 fills the room with axes-aligned bounding boxes that give false hits.
[11:36] Ubit Umarov: aim is not physics
[11:36] Robert Adams: I was wondering if there could be a use for new ossl functions just for weapons... we're trying to using general physics but maybe additional functions (like for vehicles) would make them work better
[11:36] Andrew.Hellershanks oh, of course.
[11:37] Ubit Umarov: ( until a bullet is rez.. if used)
[11:37] Kayaker.Magic no rezzing here
[11:37] Ubit Umarov: general physics are ok of a bullet
[11:37] Kayaker.Magic well, not for me.
[11:37] Ubit Umarov: doing it another way is code replication
[11:37] Andrew.Hellershanks Robert, that is an interesting idea. Someone would have to make a list of what would be needed to make better weapons then someone would have to do the coding.
[11:38] Ubit Umarov: Andrew i just said is not a good idea lol
[11:38] Robert Adams: the Bullet physics engine has per object collision margins... it could change the collision margins based on size and speed of the object but it would be easier if it was told that the object was a bullet
[11:38] Ubit Umarov: the issue with physics is the dcd
[11:39] Andrew.Hellershanks Kayaker listed 13 open mantis reports related to weapons during his talk at the OSCC. Getting those addressed would be the first place to start before adding something new.
[11:39] Robert Adams: are the mantis reports listed in his slides? (/me couldn't attend but would be interested in the list)
[11:39] Ubit Umarov: did you even tested the bullet ccd robert ?
[11:40] Andrew.Hellershanks Kayaker, I know what you mean. I have a snowball thrower where the snowball bounces off what it hits
[11:40] Ubit Umarov: ode does not have ccd :(
[11:40] Kayaker.Magic Turns out a lot of my testing was done on Kitely OpenSim 0.8.2 and at least one of those mantises was fixed in 0.9. I'm re-testing them all today.
[11:40] Andrew.Hellershanks Ubit, Would you mind explaining ccd and dcd for those (not) here who may not know those initials.
[11:41] James.atLLOUD ty Andrew
[11:41] Robert Adams: I did test it ubit but found it pretty CPU intensive
[11:41] Ubit Umarov: dcd it test collision.. move.. test collision -- move
[11:41] Ubit Umarov: once each 20ms
[11:42] Ubit Umarov: ccd is use of some technics to pretend that check is continus in time
[11:42] Kayaker.Magic 50 times per second!
[11:42] Ubit Umarov: but SL havok as ccd and ll had it off also doing dct
[11:42] Andrew.Hellershanks IIRC, cd is collision detection. The first letter is continuous or descrete.
[11:42] Ubit Umarov: bc it is heavy
[11:43] Robert Adams: think of two meshes coming together... all the comparisons to see if the vertices define overlapping objects... takes CPU
[11:43] James.atLLOUD ah, makes sense
[11:43] Ubit Umarov: now a check each 20ms is 20cm move at 10m/s
[11:43] Robert Adams: if things are moving, though, the sampling for overlap might happen at the wrong time
[11:43] Ubit Umarov: and a 10m/s is a slow bullet with a range of a few meters
[11:44] Robert Adams: so Continuious Collision Detection (CCD) computes possible collisions before they happen so the collisions can be properly reported
[11:44] Robert Adams: as you can imagine, that takes even more CPU but it gets collisions correct more often
[11:45] Ubit Umarov: well all engines cheat on the continuous part ;)
[11:45] Robert Adams: if BulletSim knew an object was a colliding bullet, it could turn on CCD selectively... but, for the moment, it's turned off for the general case
[11:45] Ubit Umarov: but is heavy we can't use it ( and ode doesn't have it )
[11:46] Ubit Umarov: you can do it per object on bullet ?
[11:47] Kayaker.Magic Hmm, but a possible OSSL function would be one a bullet could use to tell the physics engine to turn on CCD
[11:47] Ubit Umarov: i don't have ccd on ode :p
[11:47] Robert Adams: that's the sort of thing that would be possible... but, as UBit says, we'd have to figure out how it adapts in all the physics engine
[11:48] Robert Adams: or at least the two we have hackers for :)
[11:48] Ubit Umarov: :)
[11:49] Ubit Umarov: we could also add a balist path check, but bullet doesn't do ray casts ( just to get even on ode not having cc )
[11:50] Robert Adams hangs head over being lazy and not adding physics engine raycast to BulletSim
[11:50] Ubit Umarov: :)
[11:50] Robert Adams: Bullet can do it, I just haven't done all the plumbing to use the physics engine
[11:51] Ubit Umarov: i would like to delete that rayV3 thing out of lslapi.cs
[11:51] Andrew Hellershanks: Make it a new years resolution to do some plumbing. :)
[11:52] Robert Adams: LOL...
[11:52] Andrew Hellershanks: What is the difference between V1 and V3 ray casting?
[11:52] Ubit Umarov: v1 is just bounding boxes thing
[11:52] Kayaker.Magic Delete rayV3? you mean the option to select llCastRay V3 in the ini file?
[11:53] Ubit Umarov: no the CODE also :p
[11:53] Kayaker.Magic But it is still a good option when you are using Bullet.
[11:54] Ubit Umarov: yes thats why i said i "would like"
[11:54] Ubit Umarov: can't do it bf bullet doing it
[11:55] Kayaker.Magic I see
[11:55] Ubit Umarov: and rays are used elsewhere
[11:55] Ubit Umarov: like sits
[11:55] Ubit Umarov: and camera occultation checks
[11:56] Ubit Umarov: ( well when a prim gets btw camera and avatar on default camera setting )
[11:58] Ubit Umarov: hmm im not checking terrain on that.. ll does
[11:58] Ubit Umarov: but backside of terrain is transparent on viewers so...
[12:01] Ubit Umarov: silence...
[12:01] Andrew Hellershanks: :)
[12:01] Andrew Hellershanks: A couple of the other recent changes in OpenSim are some changes related to handling of mesh, and support for SSL.
[12:01] Lucy Afarensis: It has been interesting
[12:02] Ubit Umarov: errr SSL is on a test branch
[12:02] Ubit Umarov: tons of work needed
[12:02] Andrew Hellershanks: Ah, yes. I remember that being mentioned.
[12:02] Robert Adams: Ubit has been busy
[12:02] Ubit Umarov: a potencial big problem with money modules
[12:02] Andrew Hellershanks: If anyone is interested in SSL support in OpenSim they can test out the branch.
[12:02] Andrew Hellershanks: What is the name of the branch that has updated SSL support?
[12:03] Ubit Umarov: possible not that easy to fix before moving to .net4.5
[12:03] Ubit Umarov: httptests
[12:03] Andrew Hellershanks: Anything else for today?

Personal tools
About This Wiki