Chat log from the meeting on 2024-01-09

From OpenSimulator

Jump to: navigation, search
[12:11 PM PT]  Ubit Umarov: welcome to 2024
[12:11 PM PT]  Gavin.Hird thank you
[12:11 PM PT]  Ubit Umarov: i mean to fist meeting of 2024 :)
[12:11 PM PT]  Ubit Umarov: and first..
[12:11 PM PT]  Clifford.Hanger its gonna get rough?
[12:12 PM PT]  Ubit Umarov: we had no major code changes :(
[12:12 PM PT]  Vincent.Sylvester Not since last week anyways
[12:12 PM PT]  Gavin.Hird you pusehd something...
[12:12 PM PT]  Gavin.Hird pushed*
[12:13 PM PT]  Vincent.Sylvester Still trying to chase cats around with the pbr stuff heh
[12:14 PM PT]  Ubit Umarov: yes a minor fix for pbr on attachments and more "cosmetic" changes here and there
[12:14 PM PT]  Vincent.Sylvester Removing old code no longer in use, always helpful to make sense of the mess :)
[12:15 PM PT]  Vincent.Sylvester Guess we won't have a meeting next week if osg is down huh, unless we move somewhere else
[12:15 PM PT]  Gavin.Hird will it be down?
[12:15 PM PT]  Vincent.Sylvester I heard something about the 12th to 15th or something idk I didn't pay attention
[12:15 PM PT]  Ubit Umarov: hmnm this region eep looks pretty ugly with fs 7.1.2
[12:16 PM PT]  Gavin.Hird lol
[12:16 PM PT]  Gavin.Hird I have meeting rooms at
[12:16 PM PT]  Clifford.Hanger starts 16..00hrs on the 10th
[12:16 PM PT]  Ubit Umarov: ( comparing to 6.6.8. on my alt out there :)
[12:16 PM PT]  Gavin.Hird
[12:18 PM PT]  Vincent.Sylvester Last time moved to OpenSim region on zw, moving somewhere else might be fun, change of wallpaper :)
[12:19 PM PT]  Vincent.Sylvester I can put the change on the wiki.... for those that actually read it
[12:20 PM PT]  Ubit Umarov: yes, seems osgrid will go down tomorrow, so they can move things to a new machine
[12:20 PM PT]  Clifford.Hanger hi Andrew!
[12:20 PM PT]  Ubit Umarov: hopefuly all will be up before our next meeting
[12:20 PM PT]  Gavin.Hird Hi Andrew
[12:20 PM PT]  Andrew Hellershanks: Hello, everyone.
[12:21 PM PT]  Gavin.Hird a new machine only?
[12:21 PM PT]  Andrew Hellershanks: Sorry I'm late. I was on an important phone call with someone I have been having trouble contacting.
[12:21 PM PT]  Ubit Umarov: well it is a big machine :)
[12:22 PM PT]  Gavin.Hird remember I worked in IBM, so define big, haha
[12:22 PM PT]  Clifford.Hanger likes big machines
[12:22 PM PT]  Ubit Umarov: a amd Epyc cpu etc
[12:22 PM PT]  Ubit Umarov: ibm also made small things :)
[12:22 PM PT]  Ubit Umarov: like a thing called..  hmm ibm pc?
[12:23 PM PT]  Gavin.Hird which is why they sometimes were called Itty Bitty Machines
[12:23 PM PT]  Andrew Hellershanks: I've done a bit of programming for IBM 360 series and era computer. Big is relative.
[12:23 PM PT]  Clifford.Hanger never caught on lol
[12:23 PM PT]  Vincent.Sylvester I saw some 3d models for a 1800 and 1401 recently, haven't had the time to clean them up and import them though
[12:24 PM PT]  Andrew Hellershanks: Nice.
[12:24 PM PT]  Gavin.Hird they are on Sketchfab?
[12:24 PM PT]  Gavin.Hird I already have them
[12:24 PM PT]  Joe Magarac: There is an IBM 1401 simulator written using Unreal Engine. You can mount tapes, put cards in the card reader, toggle the switches, and it all works.
[12:24 PM PT]  Gavin.Hird no blinken-lights
[12:25 PM PT]  Gavin.Hird will it compile FORTRAN-2?
[12:25 PM PT]  Joe Magarac: I think so.
[12:25 PM PT]  Orbert.Tatham coughs into his hand
[12:25 PM PT]  Andrew Hellershanks: The blinken-lights are the best part ;)
[12:26 PM PT]  Vincent.Sylvester Fortran as script language would be fun, but probably not very useful xD
[12:26 PM PT]  Joe Magarac:
[12:26 PM PT]  Ubit Umarov: going restart this fs .. brb
[12:26 PM PT]  Andrew Hellershanks: Based on the above text I'm going to hazard a guess that I haven't missed much of any discussion related to Open Simulator?
[12:26 PM PT]  Joe Magarac: No.
[12:27 PM PT]  Andrew Hellershanks: I have a few printouts of programs that were written in Fortran.
[12:27 PM PT]  Gavin.Hird I bought an ASR 30 model, but it cannot be properly scaled down for import because the mesh in the keys gets too thin
[12:27 PM PT]  Andrew Hellershanks: Ah, Ubit is back. Let's get this gathering back on topic.
[12:27 PM PT]  Ubit Umarov: :)
[12:29 PM PT]  Andrew Hellershanks: Ubit, has there been any feedback since switching master to dotnet?
[12:29 PM PT]  Ubit Umarov: none
[12:29 PM PT]  Vincent.Sylvester We have had some things on mantis, but those were resolved
[12:29 PM PT]  Ubit Umarov: guess those that do run from git where running dotnet6 already
[12:30 PM PT]  Vincent.Sylvester Performance is up slightly that much I could tell fairly easily
[12:31 PM PT]  Andrew Hellershanks: Good to hear that there hasn't been any major issues.
[12:31 PM PT]  Vincent.Sylvester There were some complaints about the fact that scripts sometimes don't transfer, because their binaries are not directly compatible
[12:31 PM PT]  Gavin.Hird for me it is very stable
[12:31 PM PT]  Gavin.Hird have not had a single region crash at all
[12:32 PM PT]  Vincent.Sylvester As dotnet moves forward with things will become more incompatible as well, especially with pbr stuff, taking content from it back to mono you strip that data out basically
[12:32 PM PT]  Andrew Hellershanks: FYI, I might be moving the repos for the addon modules I maintain to Sourceforge due to problems arising from github now requiring two-factor authentication.
[12:33 PM PT]  Gavin.Hird yeah, that is a bit of a pain
[12:33 PM PT]  Orbert.Tatham Somebody always screws up a good thing
[12:33 PM PT]  Andrew Hellershanks: indeed
[12:33 PM PT]  Ubit Umarov: well as i said.. this region eep looks very ugly on fs 7.1.2
[12:33 PM PT]  Gavin.Hird bitbucket git still works
[12:33 PM PT]  Ubit Umarov: ll totally messed it up
[12:33 PM PT]  Vincent.Sylvester Sourceforge is not a good platform
[12:34 PM PT]  Gavin.Hird agreed
[12:34 PM PT]  Gavin.Hird I hate it
[12:34 PM PT]  Vincent.Sylvester Sourceforge was bought while back and they have had some shady things going on from time to time
[12:34 PM PT]  Gavin.Hird who bought it?
[12:35 PM PT]  Andrew Hellershanks: I know some people don't like SF but I maintain another project over there so I already have an SF account. SF also makes it easy to import projects from github. I just need the projects somewhere other than just github as a backup.
[12:35 PM PT]  Vincent.Sylvester You know you can grab a yubikey for 40 bucks and then use their auth app to generate those mfa codes
[12:36 PM PT]  Vincent.Sylvester Defeats the mfa point, but it is less hassle than keeping a phone nearby
[12:37 PM PT]  Gavin.Hird ask not what github can do for you - ask what you can buy for github
[12:37 PM PT]  Joe Magarac: Agreed. Have a non-Github backup.
[12:37 PM PT]  Andrew Hellershanks: I was given a Yubikey to see if I could make use of it to secure a program I work on. It would be one more program to install on both my desktop and my laptop. After that I would need a way to keep track of where the Yubikey is located for when I need it.
[12:37 PM PT]  Orbert.Tatham Anybody be willing to divide up the cost of our own "Github" equivalent?
[12:37 PM PT]  Joe Magarac: Although, unlike Google, Microsoft tends not to discontinue B2B products on short notice.
[12:38 PM PT]  Gavin.Hird true
[12:38 PM PT]  Vincent.Sylvester Gitlab is free, just needs someone familiar with it to maintain it, cause it needs regular updates and sometimes likes to break itself
[12:38 PM PT]  Joe Magarac: The risk with Github is being banned.
[12:39 PM PT]  Vincent.Sylvester No idea why we are not putting the search part into core anyways.
[12:39 PM PT]  Andrew Hellershanks: Github is supposed to handle OTP but after getting it set up it seems they don't handle OTP. When it wants me to authenticate it doesn't show me the string of characters I need to feed to the oathtool that will generate the code github wants.
[12:39 PM PT]  Gavin.Hird I have Kalithea running for myself, but the developer stopped applying changes 7 months ago, and it only runs with Python 3.8 so I guess that is toast too
[12:39 PM PT]  Vincent.Sylvester Apart from the webroot things the module itself hasn't changed at all and I don't think many use a different module for search if there even exists one
[12:40 PM PT]  Gavin.Hird
[12:40 PM PT]  Andrew Hellershanks: Vincent, Which search part?
[12:40 PM PT]  Vincent.Sylvester The region module
[12:41 PM PT]  Andrew Hellershanks: Gavin, first time I'm hearing of Kallithea.
[12:41 PM PT]  Andrew Hellershanks: It looks interesting
[12:41 PM PT]  Gavin.Hird it is not particularly well know. It handles Mercurial repos too
[12:42 PM PT]  Andrew Hellershanks: Gavin, I noticed the hg extensions on a couple of files.
[12:42 PM PT]  Gavin.Hird yep, I have all the old HG repos that were zaped from Bitbucket
[12:47 PM PT]  Vincent.Sylvester Joe, didn't you have a question last week about how crossings happen in the viewer, maybe Gavin can provide some info on that
[12:48 PM PT]  Joe Magarac: OK.
[12:48 PM PT]  Clifford.Hanger oh sharpview?
[12:48 PM PT]  Joe Magarac: I've been putting more instrumentation into Sharpview for region crossings.
[12:48 PM PT]  Joe Magarac: I can watch them fail in SL.
[12:48 PM PT]  Clifford.Hanger is there anyway to add a grid to it?
[12:48 PM PT]  Joe Magarac: In SL, single crossings without vehicles always work.
[12:48 PM PT]  Joe Magarac: ?
[12:48 PM PT]  Joe Magarac: Grid?
[12:49 PM PT]  Gavin.Hird but not with vehicles?
[12:49 PM PT]  Clifford.Hanger gives a choice of 4
[12:49 PM PT]  Joe Magarac: Double region crossings with vehicles sometimes fail.
[12:49 PM PT]  Joe Magarac: OS is not doing as well.
[12:49 PM PT]  Joe Magarac: I need to test more.
[12:49 PM PT]  Joe Magarac: But it will have to wait  until next month because I need to do some other things first.
[12:49 PM PT]  Gavin.Hird does the vehicle fail on physics?
[12:50 PM PT]  Joe Magarac: Like add sitting.
[12:50 PM PT]  Gavin.Hird in the viewer I mean
[12:50 PM PT]  Joe Magarac: I have not tried a vehicle in OS yet.
[12:50 PM PT]  Clifford.Hanger i need to get to my next assignation
[12:50 PM PT]  Vincent.Sylvester If I recall you are trying to move diagonally as well into the corner region, which can be problematic
[12:50 PM PT]  Clifford.Hanger thanks everyone
[12:50 PM PT]  Joe Magarac: Yes. But it should work.
[12:50 PM PT]  Ubit Umarov: fast crossings will most likely blow up...
[12:50 PM PT]  Gavin.Hird Most TPV (registered by LL) have access to teh Havok physics
[12:50 PM PT]  Joe Magarac: the question is, can we fix that?
[12:51 PM PT]  Gavin.Hird if not you have to use the one that the Alchemy guy wrote for opensim
[12:51 PM PT]  Joe Magarac: I will have more to say about this in a few weeks.
[12:51 PM PT]  Gavin.Hird otherwise you will not get a vehicle to cross
[12:51 PM PT]  Vincent.Sylvester Reason I bring it up, because after the meeting last week a thought got into my head about this
[12:51 PM PT]  Joe Magarac: Havok doesn't do much in the viewer.
[12:51 PM PT]  Joe Magarac: Mostly used for mesh uploads.
[12:51 PM PT]  Ubit Umarov: vavoc is just mesh tools
[12:51 PM PT]  Gavin.Hird it also works on region crossings
[12:52 PM PT]  Ubit Umarov: havoc...
[12:52 PM PT]  Vincent.Sylvester When you move diagonally you cross into the region to the side and then into the corner region most likely. That results in two crossings back to back in very quick succession that can cause issues
[12:52 PM PT]  Joe Magarac: Yes.
[12:52 PM PT]  Ubit Umarov: replaced by hadc(?) on viwers for opensim
[12:52 PM PT]  Joe Magarac: But  the viewer protocol can handle that.
[12:52 PM PT]  Vincent.Sylvester So it is an edge case, literally, of skipping that first crossing if we are too close right, ideally instead of crossing to the side first we'd just jump ahead and cross into the corner region directly
[12:53 PM PT]  Vincent.Sylvester Say you are within a few meters for a region corner, going fast enough, figure out which region to cross on the corner, send the user there instead
[12:53 PM PT]  Ubit Umarov: as i said fast crossings in succession will most likely blowup
[12:53 PM PT]  Joe Magarac: All crossings are across edges. There are race conditions when two crossings happen in rapid succession.
[12:53 PM PT]  Vincent.Sylvester That would require some logic on our end, but could "redirect" the crossing
[12:53 PM PT]  Ubit Umarov: like tps
[12:53 PM PT]  Joe Magarac: Yes, they blow up, but they shouldn't.
[12:53 PM PT]  Joe Magarac: It's a bug.
[12:53 PM PT]  Joe Magarac: And I want it fixed.
[12:54 PM PT]  Vincent.Sylvester On OpenSim they will blow up because of how crossings work internally, we just can't move the data quick enough
[12:54 PM PT]  Ubit Umarov: well its from udp etc
[12:54 PM PT]  Vincent.Sylvester So I been thinking, why try, just skip right to the diagonal region instead
[12:55 PM PT]  Joe Magarac: Because regions only talk to neighbors across edges. They don't talk to diagonal regions.
[12:55 PM PT]  Orbert.Tatham I have had cases on region crossings where I was too close to parallel to the edge, and it hung up the vehicle
[12:56 PM PT]  Andrew Hellershanks: If the X/Y says the person is very close to a corner they may be passing through but there is no guarantee that is the case. Assuming they are passing through could lead to another "bug".
[12:56 PM PT]  Orbert.Tatham Moving along the edge by accident
[12:56 PM PT]  Ubit Umarov: btw you tested unittest1/ubittest2 ?
[12:56 PM PT]  Ubit Umarov: (ubit.. not unit...)
[12:56 PM PT]  Joe Magarac: Actually, region crossings trigger when you are 1m past an edge, to avoid frantic toggling when on top of an edge.
[12:57 PM PT]  Joe Magarac: Ubit, haven't gotten to testing on your test grids yet.
[12:57 PM PT]  Joe Magarac: But I will.
[12:57 PM PT]  Ubit Umarov: not opensim.. ( currently )
[12:57 PM PT]  Joe Magarac: Please send me a notecard with the info I need to use those. Thanks.
[12:57 PM PT]  Vincent.Sylvester If the next position would put you into the diagonal region it does cross you there, but you have to hit the corner exactly and be going fast enough
[12:58 PM PT]  Ubit Umarov: what info?
[12:58 PM PT]  Orbert.Tatham your test servers, UBit
[12:58 PM PT]  Ubit Umarov: tp to one of them and test crossing to the other..
[12:58 PM PT]  Ubit Umarov: they are here
[12:58 PM PT]  Joe Magarac: On OSgrid? I've used those. OK.
[12:58 PM PT]  Ubit Umarov: in fast jsut north from this
[12:58 PM PT]  Ubit Umarov: i fact..
[12:59 PM PT]  Ubit Umarov: you can see them on map here
[12:59 PM PT]  Joe Magarac: The big region with some PBR objects on a plateau.
[12:59 PM PT]  Orbert.Tatham I see them
[12:59 PM PT]  Orbert.Tatham Thank you
[12:59 PM PT]  Vincent.Sylvester If you have two regions that are just connected via a corner and you yeet yourself into that corner it'll plop you out the other side, but it only works if you are pushing
[12:59 PM PT]  Ubit Umarov: well no edge crossing there
[12:59 PM PT]  Orbert.Tatham No corner crossing, right
[12:59 PM PT]  Ubit Umarov: yeah i mean no corner..
[12:59 PM PT]  Joe Magarac: Right, I need edges and corners I can walk and drive across.
[1:00 PM PT]  Orbert.Tatham Can you add two more?
[1:00 PM PT]  Ubit Umarov: but corner crossings will be asking for trouble anyways
[1:00 PM PT]  Joe Magarac: I found one spot on OSgrid and did some testing.
[1:00 PM PT]  Gavin.Hird most of the crossing happens in indra/newview/llviewerobject.cpp
[1:00 PM PT]  Joe Magarac: I want to get corners to work on OS so I can shame LL into fixing theirs.
[1:01 PM PT]  Andrew Hellershanks: :)
[1:01 PM PT]  Orbert.Tatham Amen to that
[1:01 PM PT]  Ubit Umarov: well issue maybe from the protocol..
[1:01 PM PT]  Ubit Umarov: how updates are sent via udp etc
[1:01 PM PT]  Joe Magarac: Viewer  Sim is OK for corners. I've been looking at that in great detail.
[1:01 PM PT]  Orbert.Tatham Each region has 8 neighbors, not four this way
[1:01 PM PT]  Ubit Umarov: well some may be udp   http
[1:02 PM PT]  Joe Magarac: all that part works, at least for SL.
[1:02 PM PT]  Joe Magarac: Gotta go, next meeting. More next week.
[1:03 PM PT]  Ubit Umarov: and opensim regions will talk to a corner region.. it if does find it is the crossing destination
[1:03 PM PT]  Andrew Hellershanks: ok, Joe. See you next week. Good luck with your project.
[1:03 PM PT]  Ubit Umarov: think is that in most cases that will be 2 crossings in fast sequence..
[1:03 PM PT]  Vincent.Sylvester The problem is if you move at say 256,254 diagonal you get placed in the region to the side first, then you instantly cross into the diagonal region. OpenSim falls over, because it cannot process these in less than a second. If you hit 256,256 exactly and move quick it'll put you into the diagonal region. More so if the only available region is the diagonal one.
[1:03 PM PT]  Orbert.Tatham Ubit, I know this may not be easy, but could you put corner regions on par with edge regions, making them the same crossing protocol?
[1:04 PM PT]  Ubit Umarov: they are on same protocol
[1:04 PM PT]  Vincent.Sylvester A potential fix would be to check the "next position" on crossing to see if it is diagonal in another region and just change the target of the crossing from the side to the diagonal region
[1:05 PM PT]  Orbert.Tatham That is more to what I had in mind, yeah
[1:05 PM PT]  Vincent.Sylvester Meaning if pos ~256,256 check direction vector, contact diagonal region for crossing
[1:05 PM PT]  Ubit Umarov: oe just put a big prim on corners "no not cross here"
[1:05 PM PT]  Ubit Umarov: :)
[1:05 PM PT]  Gavin.Hird is that fast enough to avoid rubber banding in the viewer?
[1:06 PM PT]  Gavin.Hird to make the reroute I mean
[1:06 PM PT]  Vincent.Sylvester That's what I am worried about, because the bounding box for diagonal crossing would need to depend on velocity as well
[1:06 PM PT]  Gavin.Hird yes
[1:06 PM PT]  Vincent.Sylvester Going faster it needs to be bigger to avoid side crossing first, but all that adds overhead to the crossing code
[1:06 PM PT]  Ubit Umarov: in fact a few years ago i did test corners crossings with ubode
[1:07 PM PT]  Ubit Umarov: and back then .. kinda worked,  ( on same network etc )
[1:07 PM PT]  Gavin.Hird if you are just a bit off, will not 3 regions start to compete about the agent or object
[1:07 PM PT]  Gavin.Hird actually the agent is an object
[1:08 PM PT]  Vincent.Sylvester If you crash into the corner and only have the corner region available to cross it works a lot better, it's the region to the side getting the crossing trigger that breaks it all yeah
[1:08 PM PT]  Ubit Umarov: a ugly thing was having a elastic prim jsut on the other side of crossing
[1:08 PM PT]  Ubit Umarov: sending the vehicle fast back :)
[1:08 PM PT]  Gavin.Hird did it snap back then?
[1:09 PM PT]  Ubit Umarov: well it kinda went back..
[1:09 PM PT]  Ubit Umarov: or into the void :)
[1:09 PM PT]  Gavin.Hird oh
[1:10 PM PT]  Vincent.Sylvester If you have only a region diagonally you can drive into it really fast and even if you hit it at say 256,254 it'll plop you out the other side fine, it's the calculation of "next position" that is kinda key here
[1:10 PM PT]  Gavin.Hird another thing at a diagonal crossing, we cannot assume the motion is straight lines. the agent or object may be moving on a curved path
[1:10 PM PT]  Vincent.Sylvester It tries to find where you would end up and see if there is something there to send you to
[1:11 PM PT]  Vincent.Sylvester Bounding box approach would work, but not an elegant solution
[1:11 PM PT]  Ubit Umarov: it is abotu root prim mostly
[1:12 PM PT]  Vincent.Sylvester No I mean like bounding box at each corner of the region that if you are in it and moving diagonally you get crossed diagonally instead
[1:12 PM PT]  Vincent.Sylvester Kinda like a teleporter placed at the region corner
[1:12 PM PT]  Ubit Umarov: if you drive a long car agains the edge.. its front wheels may fall :)
[1:13 PM PT]  Ubit Umarov: bc physcs only does ground a few meters outside the egde
[1:13 PM PT]  Ubit Umarov: and edge is by root prim position..
[1:13 PM PT]  Vincent.Sylvester If you throw prims at it they might drop velocity completely and fall into the void, but that still happens once in a while on normal crossings
[1:13 PM PT]  Ubit Umarov: sl does have larger margins on that
[1:14 PM PT]  Ubit Umarov: possible 16m
[1:14 PM PT]  Ubit Umarov: thing they also share terrain on those 16m..
[1:15 PM PT]  Ubit Umarov: we do not
[1:15 PM PT]  Gavin.Hird meaning?
[1:15 PM PT]  Ubit Umarov: land extends 16m outside the edge for physics,,,
[1:15 PM PT]  Vincent.Sylvester Ever tried editing terrain across region borders, fun stuff
[1:16 PM PT]  Gavin.Hird oh, ok
[1:16 PM PT]  Ubit Umarov: on a open edge it is just flat..
[1:16 PM PT]  Gavin.Hird I do it all the time
[1:16 PM PT]  Ubit Umarov: if there is a region there, then it gets the terrain the other region has there
[1:16 PM PT]  Gavin.Hird it works using the flatten terrain brush
[1:16 PM PT]  Ubit Umarov: on opensim..  there is no such share
[1:17 PM PT]  Vincent.Sylvester Isolation by design
[1:17 PM PT]  Ubit Umarov: terrain edit works on several cases..
[1:17 PM PT]  Vincent.Sylvester Cause you don't want your neighbor to mess with your stuff basically
[1:17 PM PT]  Vincent.Sylvester Which is a good thing, but also means some stuff is more glitchy
[1:18 PM PT]  Ubit Umarov: bc viewers send same brush data and global posioton to both regions
[1:18 PM PT]  Ubit Umarov: but..  they may "desync.."
[1:18 PM PT]  Gavin.Hird the only brush that works across regions are flatten
[1:19 PM PT]  Gavin.Hird but by waring the size and magnitude you can get pretty good results
[1:19 PM PT]  Vincent.Sylvester It's been driving me nuts trying to align terrain, but yeah same cause, regions not "interacting" with each other for good reason
[1:20 PM PT]  Ubit Umarov: well there are reasons why we do have large regions :)
[1:20 PM PT]  Vincent.Sylvester Regarding crossings I think we could attempt the bounding box idea, force diagonal crossing if the alternative would be a too fast side crossing, cause that's just a bandaid really
[1:20 PM PT]  Ubit Umarov: croosing had always been a fail
[1:21 PM PT]  Vincent.Sylvester Probably be a better solution for the viewer to say "hey I wanna go to this region" cause it more closely knows what the user is requesting
[1:21 PM PT]  Vincent.Sylvester We just get packets for position change and those may drop
[1:21 PM PT]  Ubit Umarov: no that is nor viewer decision
[1:22 PM PT]  Orbert.Tatham Even the view can only guess, but it is closer to the user's control inputs than the server
[1:22 PM PT]  Ubit Umarov: regions decide that
[1:22 PM PT]  Vincent.Sylvester Unfortunately it is not yeah
[1:22 PM PT]  Vincent.Sylvester Difference between crossing and teleport
[1:22 PM PT]  Gavin.Hird doesn't the region decide where the object is at any time
[1:22 PM PT]  Vincent.Sylvester Yep
[1:23 PM PT]  Vincent.Sylvester Viewer just says "move" which is udp part so with packet loss might not notice what I mean
[1:23 PM PT]  Gavin.Hird so the viewer acts on the  position reported by the region
[1:23 PM PT]  Ubit Umarov: well object teleport can do crossings
[1:23 PM PT]  Ubit Umarov: even long range tps
[1:23 PM PT]  Ubit Umarov: but a very ugly thing to watch
[1:24 PM PT]  Vincent.Sylvester Yep, viewer tells region what to do and it reports back once it did that, viewer only updates then
[1:24 PM PT]  Vincent.Sylvester Feedback loop
[1:24 PM PT]  Ubit Umarov: (joe may not knoe about objects teleports
[1:24 PM PT]  Gavin.Hird obviously there is an interaction there, but you sometimes see some ugly readjustments to what the regions sais
[1:25 PM PT]  Ubit Umarov: regions decide postions
[1:25 PM PT]  Vincent.Sylvester Cause it moves first and asks questions later xD
[1:25 PM PT]  Ubit Umarov: they do run the physics..
[1:25 PM PT]  Ubit Umarov: ( and/or scripts )
[1:26 PM PT]  Vincent.Sylvester Cause the "cannot move" is more rare than "can move" hence most checks for that happen after it is applied, that's why you rubberband on ban lines and such
[1:26 PM PT]  Vincent.Sylvester Plus all that is not exactly accurate either
[1:26 PM PT]  Vincent.Sylvester Avatars technically constantly bouncing around
[1:26 PM PT]  Ubit Umarov: oops 26min past hour :)
[1:27 PM PT]  Gavin.Hird so we drop a few frames on region crossings to minimize the visual bouncing
[1:28 PM PT]  Gavin.Hird it might be as much as 1 sec
[1:28 PM PT]  Vincent.Sylvester Can be yeah, most times is pretty fast thankfully
[1:28 PM PT]  Vincent.Sylvester If regions are properly hosted anyways
[1:29 PM PT]  Gavin.Hird I mean there is a deliberate 1 sec where no new frames are rendered on crossing
[1:29 PM PT]  Ubit Umarov: never seen that
[1:29 PM PT]  Gavin.Hird even FS does it (they came up with the idea)
[1:30 PM PT]  Ubit Umarov: objects or avatars are just where region told them to be, or do predict movement
[1:30 PM PT]  Gavin.Hird not sure if it is a full second, but a number of frames are not refreshed
[1:30 PM PT]  Ubit Umarov: well see no chage on frames..
[1:31 PM PT]  Ubit Umarov: objects on regions keep moving...
[1:31 PM PT]  Ubit Umarov: while we can get rubber band..
[1:31 PM PT]  Ubit Umarov: or stop on edge a bit
[1:31 PM PT]  Gavin.Hird the dropped frames are to minimize the apperance of rubberbanding
[1:32 PM PT]  Gavin.Hird it is seen like a small glitch
[1:32 PM PT]  Ubit Umarov: that would cause jitter on all the other moving things
[1:32 PM PT]  Gavin.Hird it can do
[1:33 PM PT]  Gavin.Hird the thing is that the user usually is focused on the avatar on crossing, so whatever is happening in the frame does not have any attention
[1:34 PM PT]  Gavin.Hird it is just local to the viewer, everything else goes not as normal
[1:37 PM PT]  Andrew Hellershanks: We are half past the hour mark. Before we go do we need to finalize a backup meeting location in case osgrid is still down this coming Tuesday?
[1:39 PM PT]  Gavin.Hird as I said I have a region with meeting rooms
[1:39 PM PT]  Andrew Hellershanks: Gavin, what is the URL? I wasn't here when that was mentioned.
[1:40 PM PT]  Gavin.Hird
[1:41 PM PT]  Gavin.Hird
[1:41 PM PT]  Andrew Hellershanks: ok. Are the meetings numbered? We can always pick one after we arrive.
[1:42 PM PT]  Orbert.Tatham Need to get home - see you all later
[1:42 PM PT]  Andrew Hellershanks: s/meetings/meeting rooms/
[1:42 PM PT]  Andrew Hellershanks: ok, Orbert. We are about done here.
[1:42 PM PT]  Gavin.Hird no, we just pick a room
[1:42 PM PT]  Andrew Hellershanks: See you next week.
[1:42 PM PT]  Orbert.Tatham Peace
[1:42 PM PT]  Andrew Hellershanks: ok, ty Gavin.
[1:42 PM PT]  Gavin.Hird see you all next week then
[1:43 PM PT]  Ubit Umarov: cya
[1:43 PM PT]  Andrew Hellershanks: We can just wait by the landing area and pick a room after people arrive. Late comers can check the map for a closely packed group of dots.
[1:44 PM PT]  Andrew Hellershanks: Thank you all for coming. See you again next week.k
Personal tools
About This Wiki