Chat log from the meeting on 2019-06-18

From OpenSimulator

Jump to: navigation, search

[11:06] Kayaker Magic: Bill, how do you get multiple colors in one llSetText in that clock?
[11:06] Sheera Khan: look at the number of prims Kay ;-)
[11:07] Andrew.Hellershanks Nice effect, Bill.
[11:07] Bill.Blight making them all line up right was not fun
[11:08] Kayaker Magic: Yes, nicely done!
[11:08] Andrew.Hellershanks I only see 2 prims.
[11:08] I'm a clock: 11:08:32 Pacific Time
[11:08] Bill.Blight If you look on the edit screen you will see that it is 4
[11:08] Bill.Blight total
[11:08] Andrew.Hellershanks Yes, I see that now.
[11:09] Gavin.Hird If I toggle displayname I see Leighton has got one
[11:10] Andrew Hellershanks: Bill, myself, dj, and Sandy have text above the names.
[11:11] Sandy Beachcomber: others have no groups active?
[11:11] Gavin.Hird there is some interaction there
[11:11] Andrew Hellershanks: The setting you want to toggle is Group titles.
[11:11] Kayaker Magic: I have a group active, and I see a few others.
[11:11] Sheera.Khan I can't switch groups here ...
[11:12] Andrew Hellershanks: Yes, Kayaker has one too. Missed mentioning him in the list.
[11:12] Sandy Beachcomber: hes on
[11:13] Sandy Beachcomber: did my text go now?
[11:14] Andrew Hellershanks: It did
[11:14] Sandy Beachcomber: now its back
[11:14] Andrew Hellershanks: It is.
[11:14] Kayaker Magic: I'm starting to get an error I never did before while uploading mesh: The viewer tells me it has triangles too small and won't even let me set physics LOD lowest.
[11:14] Gavin.Hird yes
[11:14] Sandy Beachcomber: activate group
[11:14] Sandy Beachcomber: i activated none
[11:14] Kayaker Magic: It happens on some grids but not others, on reasonable mesh like Ruth 2.0.
[11:15] Andrew Hellershanks: Kayaker, Do you have another question or three for us this week?
[11:15] Bill Blight: that is the new mesh validation in FS has nothing to do grid side
[11:15] Kayaker Magic: Then the stupid viewer won't let me select NONE again until I exit the dialong and star over.
[11:15] Ubit Umarov: i talked a bit with Beq about FS mesh validation
[11:15] Kayaker Magic: But why is it different on different grids with the same FS?
[11:15] Ubit Umarov: she may disable it for opensim
[11:16] Bill Blight: there is a debug setting to turn it off
[11:16] Kayaker Magic: And I have an old version of FS, 6.01 (56538)
[11:16] Ubit Umarov: thats outdated
[11:16] Bill Blight: FSIgnoreClientsideMeshValidation
[11:16] Bill Blight: you can't turn it off in 6.01
[11:17] Bill Blight: 6.0.2 lets you turn it off
[11:17] Bill Blight: via debug
[11:17] Bill Blight: They added the mesh validation in the 6 series
[11:18] Ubit Umarov: and some if it seems impossible to disable
[11:18] Andrew Hellershanks: I'm on 6.0.2. I haven't checked if there is a newer version available.
[11:18] Kayaker Magic: There is a tutorial on how to upgrade FS and keep all your logs and shit. The video says it's easy, just ... and then it goes on an on for 30 minutes with a dozen different things you have to do.....
[11:18] Ubit Umarov: well those validation rules are good
[11:18] Gavin.Hird I agree Ubit
[11:18] Bill Blight: just install it over the top, should keep all your logs
[11:19] Bill Blight: just don't uninstall the older first
[11:19] Ubit Umarov: a goot try to reduce the amount of bad meshes in worlds
[11:19] Bill Blight: Yeah , personally I like the validation
[11:19] Ubit Umarov: but even the ( bad) ones i used to test uploads, don't upload :)
[11:22] dj phil: did anyone here ever play with OSSL inertia functions ?
[11:22] Arielle Popstar: isnt that the point?
[11:22] Ubit Umarov: silent moment?
[11:22] Andrew Hellershanks: dj, IIRC, I tried ages ago but they didn't work then. I haven't tried recently.
[11:22] Bill Blight: I have messed with them some, gets out of control in a hurry, lol
[11:22] Gavin.Hird I am watching a build progressing...
[11:23] Ubit Umarov: you do need to keep inertia with reason of things get wild ( in a bad way )
[11:24] dj phil: Adrew : What is a good use for inertia ?
[11:24] Gavin.Hird in Xcode a full viewer rebuild takes about 12 minutes. In Visual Studio 3+ hours
[11:24] Bill Blight: I've used them to change the center of mass of some of my boats
[11:24] Andrew Hellershanks: dj: Not sure I can think of a good example at the moment.
[11:26] Ubit Umarov: dj if you need more help, that they are not for you :p
[11:26] Bill Blight: Example , root prim does not have to be center of the the vehicle/physcial object, you can offset the center of mass so the mass calculations for physical movement are based on that new center ..
[11:26] dj phil: Ubit null response
[11:26] Ubit Umarov: honest response :p
[11:26] Sandy Beachcomber: how is that inertia?
[11:27] dj phil: I come here today expressly for this
[11:27] dj phil: ok ty Ubit
[11:27] Ubit Umarov: the inertia functions do include total mass and center of mass override
[11:28] Bill Blight: when a vehicle turns or speed increass or decrease the center of mass will be the new one, so the pivot point for it's mass is calculated on the new location ..
[11:28] Ubit Umarov: their goal is to replace the estimations based on the prims, that in same cases can be bad
[11:29] Bill Blight: inertia will cause a vehicle to pivot on it's center of gravity based on those calcuations
[11:29] Ubit Umarov: well you change it in run time and things can go badly :p
[11:29] Bill Blight: for sure
[11:29] Bill Blight: been there done that
[11:29] Ubit Umarov: idea is to use during prim rez/build
[11:30] dj phil: You said could be bad and therefore could cause undesirable physical movement
[11:30] Kayaker Magic: Then it beccomes a prim parameter that persists?
[11:30] Ubit Umarov: change duing simulation may go worse, bc physics engine do not do proper conservation on angular and linear moments
[11:30] Bill Blight: for sure, changing the mass of the object in motion or the center of mass could cause it to go off in a totally un planned direction
[11:31] Ubit Umarov: well it does happen in rl a lot
[11:31] Kayaker Magic: That could be handy in RL, make getting stuff to orbit cheaper.
[11:31] Ubit Umarov: but mb a bit 2 much for the engines
[11:32] Bill Blight: be like the load shifting on a bit cargo ship , and the thing leans to the side
[11:32] Ubit Umarov: the spinnig dancer is a example
[11:32] Ubit Umarov: how the open or close arms and legs to control omega
[11:32] Sandy Beachcomber: what are trying to achive dj?
[11:33] Ubit Umarov: but that may gobadly in current engines, as i said
[11:33] dj phil: nothing right now Sandy, I'm only informing myself
[11:34] Sandy Beachcomber: cool
[11:35] dj phil: And how do you calculate this new mass center point, it's a precise formula I suppose?
[11:35] Andrew Hellershanks: precise? :)
[11:35] dj phil: predefined
[11:35] Ubit Umarov: as i said goal was to allow it to be set independently of the prims ( and their density) bc on some, meshes specially, the estimation can go badly
[11:35] Bill Blight: with the functions , you can set it to what you want, even if it is bad, ugly and dangerious
[11:36] Ubit Umarov: and there aren't that many things that do depend on it
[11:36] Ubit Umarov: err and not sure if usefull at all with bullet
[11:36] Bill Blight: there is a function to get the actual calculations so you can use that as a starting poin
[11:36] Bill Blight: and it does not work on bullet
[11:36] Ubit Umarov: since bullet places all the mass, hmm on root prim ???
[11:37] Bill Blight: yeah center of root prim
[11:37] Bill Blight: or centered on the root prim I should say
[11:37] dj phil: I understand that I can adjust with these functions
but the values are probably values that are calculated
according to physical rules ...
I should not put random numbers in there
[11:37] Ubit Umarov: well ubode does estimate center of mass and does place it there..
[11:38] Bill Blight: As I said you can get the calculated numbers and use them as the basis for your input numbers
[11:39] Ubit Umarov: you may had noticed, that "free" physical objects do rotate on ubode diferently for bullet
[11:39] Ubit Umarov: center of rotation a major dif
[11:40] dj phil: ok ubode calculates the center
some objects not good
inertia override the values ...
how to calculate good values?
[11:40] Ubit Umarov: ops
[11:41] dj phil: i can understand what do inertia
[11:41] Ubit Umarov: hmm hard example lol
[11:41] dj phil: but no how to fix with good values if some are wrong ...
[11:41] Bill Blight: the values are going to depend on what type of object you are trying to calculate for, a floating feather would be different than a lead brick
[11:42] testInertiaOvr: Script running
[11:42] dj phil: ok so initially we could say the center of mass does not suit me so I change it
[11:43] dj phil: we should not say it's bad, as if it was miscalculated
[11:43] dj phil: true ?
[11:43] Sandy Beachcomber: sounds like many experiments needed
[11:43] dj phil: changing center of mass is good
[11:43] dj phil: changing pivot point is good too ?
[11:44] dj phil: we can ? :(
[11:44] Bill Blight: This is going to be more evident on mesh objects that are not uniform in shape, but standard calculations would put the center of mass on the center of the object, where visible one end may need to be more dense to be in line with the physical look
[11:44] Ubit Umarov: see the center of rotation of this 2 prims?
[11:44] Bill Blight: is that target omega, does not apply
[11:44] dj phil: or a single prim like a doors, yes ubit
[11:44] dj phil: the pivot point
[11:44] dj phil: osPivotPoint
[11:45] Ubit Umarov: in this case they rotate around their center
[11:45] dj phil: yep
[11:45] Ubit Umarov: ( they are physical with boyance
[11:45] Ubit Umarov: on touch, the script applies rot impulse
[11:46] dj phil: yes, i don't talk for inertia objects, but for non physical object
[11:46] Ubit Umarov: bullet does not rotate like this
[11:46] Ubit Umarov: inertia only matters on physical objects
[11:46] dj phil: true
[11:46] Bill Blight: non-physical have no mass , so these functions don't apply
[11:46] dj phil: i'm ok with inertia now
[11:46] dj phil: just need to do self tests now
[11:47] Andrew Hellershanks: It doesn't apply inertia to the side of the object touched. It is rotating always in the same direction.
[11:47] dj phil: pivot point is another subject :)
[11:47] Ubit Umarov: now you should see the center closer to larger prim
[11:47] dj phil: PS: Bill i know how to create "fake" pivot"
[11:47] dj phil: i ask for pivot function :)
[11:47] Ubit Umarov: notice it?
[11:48] dj phil: can you send me this exemple ?
[11:48] Ubit Umarov: ( touch may also cause drag )
[11:48] Ubit Umarov: this is basic and not using ossl inertia
[11:48] Frigid Cryotank: Touch causes drag?
[11:48] Ubit Umarov: ubode always did this
[11:48] dj phil: is better is i can look inside on ly region
[11:49] Ubit Umarov: touch can drag the object a bit
[11:49] dj phil: interessing :)
[11:49] Ubit Umarov: seem me draging ? :)
[11:49] dj phil: can you make it copy/mod please ?
[11:50] Ubit Umarov: on single prim it also is different
[11:50] Andrew Hellershanks: We have about 10 minutes left in the hour. Before we continue talking about inertial does any one else have a question?
[11:50] Andrew Hellershanks looks at Kayaker
[11:50] Sandy Beachcomber: viewer crashes on region change and tp
[11:50] Ubit Umarov: well hard to see :9
[11:51] Andrew Hellershanks: Go ahead, Sandy.
[11:51] Sandy Beachcomber: and pointers?
[11:51] Sandy Beachcomber: any pointers
[11:51] Ubit Umarov: notice the rotation center on this cut prim ?
[11:51] dj phil: yes
[11:51] Sandy Beachcomber: any viewers better than others
[11:51] Bill Blight: you have more than one region per side of the region you are going to Sandy?
[11:51] Andrew Hellershanks: Which viewer are you using, Sandy?
[11:52] Sandy Beachcomber: in places yes
[11:52] Sandy Beachcomber: fs
[11:52] Bill Blight: that be it
[11:52] Sandy Beachcomber: 6.02
[11:52] Bill Blight: you can only have one region in view per region side, regardless of the size of the region
[11:52] Ubit Umarov: see how it is not the prim position, but more or less its actual center?
[11:52] Kayaker Magic: Here's another question: The CtrlAltStudio guy had a version of the viewer that allowed changing the parameters of an animation after uploading. So you could change animation frame rate to run fast or slow, perhaps other parameters. The question is: Doesn't that require a change to the server to send the parameters? Is there a OSSL call to send generic parameters to the viewer?
[11:52] Bill Blight: this is a confirmed viewer issue
[11:52] Sandy Beachcomber: ok bill ty
[11:53] Sandy Beachcomber: so as grid should have all regions same size ideally?
[11:53] dj phil: Hard to see ... and I do not see the numbers in the script to help me see lol
[11:53] Sandy Beachcomber: a grid
[11:53] dj phil: But i imagine ... :)
[11:53] testInertiaOvr: Script running
[11:53] Gavin.Hird if they touch Sandy
[11:54] Sandy Beachcomber: they have to for region crossings
[11:54] Bill Blight: same size or only one per side ... We spent quite a bit of time chasing that issue, only to have it confirmed as a viewer side issue. Viewers being based on SL only expect one region per side ..
[11:54] Sandy Beachcomber: all viewers?
[11:54] Bill Blight: and may get confused as to what region you are going to ..
[11:55] Andrew Hellershanks: Kayaker, that sounds like a change you would have to make before uploading to a grid. I once needed to make an animation uploaded to the grid loop and I modified the asset in the database table. That isn't something I would do on a regular pbasis. I don't think you can do much to modify an animation via scripting.
[11:55] Bill Blight: all viewers should suffer from it, but some crash faster than others
[11:55] Sandy Beachcomber: lol
[11:55] Sandy Beachcomber: crash faster
[11:55] Ubit Umarov: ok the prim is not outside with a unrealistic center of mass
[11:55] Ubit Umarov: is now
[11:55] Frigid Cryotank: <heh> Should be easier then to find the bug.
[11:55] Kayaker Magic: Discovery Grid has multiple different var sizes and bumps them up against each other in a few places. Even two 4x4's on the same side of an 8x8. I cringed but sailed a boat around the intersection multiple crossings with no problems.
[11:55] Ubit Umarov: see it?
[11:55] Sandy Beachcomber: interesting Kay
[11:56] Andrew Hellershanks: Kayaker, the boat didn't sink? :)
[11:56] dj phil: ha yes, better to se it there
[11:56] Sandy Beachcomber: any idea what code/branch?
[11:56] Ubit Umarov: now it does rotate around a irrealistic center of mass set with ossl..
[11:56] dj phil: The unrealistic center of mass is a sort of pivot point based on mass ?
[11:56] Andrew Hellershanks thinks Ubit is having too much fun with his toys. :)
[11:57] Sandy Beachcomber: lol
[11:57] Bill Blight: you could go months without it crashing , then due to , content, cache, internet, whatever it will start crashing consistently
[11:57] Bill Blight: also has a lot to do with view distance
[11:57] Kayaker Magic: Discovery was using a year old version of master at the time (from January 2018). They switched to Nani and I should go test that intersection again.
[11:57] Ubit Umarov: dj. first you need to understand a bit of physics
[11:57] Frigid Cryotank knows UBit it having too much fun with his toys.
[11:57] Ubit Umarov: if not, don't use this
[11:57] Andrew Hellershanks: Nani?
[11:57] Sandy Beachcomber: i use nani too
[11:57] dj phil: Hmmm :s
[11:58] Bill Blight: the one with all the throttles removed
[11:58] Kayaker Magic: Nani is a branch of OpenSim made by LaNani
[11:58] Andrew Hellershanks: ah, ok.
[11:58] Ubit Umarov: a region can only have one other region per each of its sides
[11:58] Arielle Popstar: more then just throttles removed
[11:58] Ubit Umarov: that is a viewers limitation
[11:58] Andrew Hellershanks: Not surprising someone went and took out all the throttles. Wonder what the grid performance is like when a lot of people are using it at the same time.
[11:59] Ubit Umarov: you can't put 2 regions on same side ONLY ONE
[11:59] Kayaker Magic: Nani also has a bunch of optimizations added, not just throttles removed.
[11:59] Sandy Beachcomber: never seen any issues on her regions with parties
[11:59] Ubit Umarov: Kayeker this meeting is about opensim core
[12:00] Kayaker Magic: People who party a lot LOVE Nani because mesh avatars load fast.
[12:00] Sandy Beachcomber: is it about core?
[12:00] Gavin.Hird problem with the Nani code she does not publish it in a way you can see diffs and commits
[12:00] Ubit Umarov: several of those so called optimizations, at the time, where just bad
[12:00] Sandy Beachcomber: or development
[12:00] Ubit Umarov: like no udp thottling
[12:00] Arielle Popstar: sometimes have to compare to be able to see better solutions
[12:01] Sandy Beachcomber: this meeting is only for core?
[12:01] Andrew Hellershanks: It wouldn't be the first time someone forked the OS code and kept their changes private.
[12:01] Sandy Beachcomber: or for development?
[12:01] Ubit Umarov: so lets stay withing main opensim code, that we can speak about
[12:01] Kayaker Magic: Back to OpenSim: Discovery Grid also has offset var regions: two 8x8 vars meeting in the middle of a third 8x8. This decreases the total number of neighboring regions to decrease the stuff the viewer had to load.
[12:01] Arielle Popstar: Nani does list her changes but just not in the usual fashion
[12:01] Sandy Beachcomber: she doesnt keep them private
[12:02] Gavin.Hird exactly Arielle
[12:02] Ubit Umarov: kay that may confuse viewers badly
[12:02] Andrew Hellershanks: If the changes made to Nani aren't private it might be interesting to see them if there are changes that could, or should, be added to core.
[12:03] Sandy Beachcomber: so Discovery fixed the "viewer "issue?
[12:03] Ubit Umarov: dunno, do they have own viewer ?
[12:03] Kayaker Magic: No, they just never saw any viewer issuess. (Yet?)
[12:03] Sandy Beachcomber: they are not private
[12:03] Kayaker Magic: I use FS there.
[12:03] Gavin.Hird do they publish the grid code?
[12:03] Bill Blight: you could go months and never have the issue
[12:03] Sandy Beachcomber: yes
[12:03] Ubit Umarov: well as i said, ther can only be one neibour per each region side
[12:04] Kayaker Magic: Like I said, Discovery was using a version of master when I last tested this.
[12:04] Ubit Umarov: thats because viewers have a fixed array size and indexes math on several code paths
[12:04] Sandy Beachcomber:
[12:04] Ubit Umarov: a typical issue it when we tp out they crash
[12:05] Sandy Beachcomber: i get that too
[12:05] Kayaker Magic: Sandy! You are not allowed to talk about that unmentionable fork of OpenSim here!
[12:05] Sandy Beachcomber: so eject me
[12:05] Gavin.Hird All changed files that deviate from the original Opensim master:
[12:05] Sandy Beachcomber: lol
[12:05] Kayaker Magic: Ubit: by "they crash" do you mean the regions or the viewer?
[12:05] Ubit Umarov: the viewer
[12:05] Gavin.Hird this is a nightmare to go through if the single commits are not publsined
[12:05] Sandy Beachcomber: all we all want is better stabler code?
[12:06] Ubit Umarov: it tries to delete what he things its the came region ( on same array location )
[12:06] Ubit Umarov: came? same !!
[12:06] Sandy Beachcomber: ok but fs is supposed top support os?
[12:06] Ubit Umarov: so again.. a region can only have ONE other region per side no matter its size
[12:07] Andrew Hellershanks: I have seen people violate that principle.
[12:07] Andrew Hellershanks: I've seen two 1x1s up against the same side of a 4x4 with a space between the two 1x1.s
[12:08] Andrew Hellershanks: I'm surprised it didn't cause instant crash of a viewer.
[12:08] Ubit Umarov: it does not... viewer things
[12:08] Sandy Beachcomber: i have much worse
[12:08] Ubit Umarov: almost certain is crash on tp out
[12:09] Sandy Beachcomber: but to be fair to ubit i do see viewer crashes
[12:09] Ubit Umarov: or when viewer closes regions
[12:09] Bill Blight: closes the wrong region to be precise
[12:09] Bill Blight: LOL
[12:09] Sandy Beachcomber: but i live in hope that the viwer crashes can be sorted?
[12:09] Ubit Umarov: viewer side thing
[12:10] Ubit Umarov: and not a simple code change rally
[12:10] Ubit Umarov: .. really
[12:10] Gavin.Hird it requires a heck of a lot of testing too
[12:10] Gavin.Hird you have to have test setups that covers most eventualities
[12:10] Ubit Umarov: so better keep it as a rule, bc viewers will hardly change that in near future
[12:10] Sandy Beachcomber: but if it aint solved it limits the use of var regions
[12:11] Ubit Umarov: ( also not needed for sl.. so.. )
[12:11] Sandy Beachcomber: so to slove it all we should stop using vars?
[12:12] Kayaker Magic: Stop putting more than one var against the side of another var.
[12:12] Gavin.Hird you can still use vars within the stated limits
[12:12] Bill Blight: Leave empty space between them
[12:12] Ubit Umarov: what part of what i said means stop using var regions ???
[12:12] Bill Blight: you can put a var next to a var, just one region per side .. a var is ONE region
[12:13] Sandy Beachcomber has to rethink new grid
[12:13] Sandy Beachcomber: wanted to sail and fly across borders
[12:14] Kayaker Magic: You can sail and fly across borders.
[12:14] Andrew Hellershanks: You can do that.
[12:14] Bill Blight: then just make them all same size vars
[12:14] Leighton.Marjoram you can have any size var you want without sim borders. Opensim doesnt have that limitation
[12:14] Sandy Beachcomber: as i sais earlier
[12:14] Ubit Umarov: well not that long ago bc of opensim code they could not have any neibours at all
[12:14] Arielle Popstar: butt the northwest corner up against the southeast corner of the var above
[12:15] Arielle Popstar: same rule as we had with mega regions
[12:15] Kayaker Magic: And arrange the vars so there is never more than 2 wherever they but up against each other on one side. Yeah, plus the diagonal ones.
[12:15] Ubit Umarov: it is not related to mega regions
[12:15] Arielle Popstar: worked the last time i tried that
[12:15] Sandy Beachcomber: sorry i wasnt aware of that "rule"
[12:15] Ubit Umarov: it is viewer side limitation, im not going to repeat that again
[12:15] Bill Blight: corners count as sides, as far as the viewer is concerned , so you can put as many around and together as you want as long as there is only one per side you can have 8 vars around the outside of a single var
[12:16] Sandy Beachcomber: ok redesign coming up
[12:16] Kayaker Magic: First I saw someone put 2 4x4 up against an 8x8, my hair stood on end. I never expected that to work!
[12:16] Bill Blight: (although the viewers see corners and corners not sides
[12:17] Ubit Umarov: ] Bill Blight: (although the viewers see corners AS corners not sides
[12:17] Bill Blight: :P
[12:17] Kayaker Magic: One 4x4 up against an 8x8, that might work, but not 2!
[12:17] Andrew Hellershanks: If you break it you get to keep the pieces. :)
[12:17] Kayaker Magic: LOL
[12:19] Andrew Hellershanks: It is getting on to twenty past now. Any last minute question or topic before we wrap up todays meeting?
[12:19] Kayaker Magic: And FYI, I have seen one 4x4 against an 8x8 and crossings worked.
[12:19] Sandy Beachcomber: lol kayaker!
[12:19] Bill Blight: yes, that is because ONE PER SIDE
[12:19] Kayaker Magic: But don't know if it ticks off the viewer long term.
[12:19] Andrew Hellershanks: Kayaker, it might work for one person but not for another depending on which viewer is in use.
[12:19] Bill Blight: no it doesn't , that still fits in the ONE PER SIDE
[12:20] Bill Blight: a var is ONE region regardless of size
[12:20] Sandy Beachcomber: its ok i relay the grid tomorrow
[12:21] Kayaker Magic: IT IS A WRAP! I say we adjourn the meeting.
[12:21] Leighton.Marjoram ok I have done math now this is serious, you could have a 10x10 var region with 8 10x10s surrounding (one per side) thats 900 standard 256m region (230,400sqm) I am not sure how much more space you would need.
[12:21] Andrew Hellershanks: I'm about to call this meeting closed and I see four people typing.
[12:21] Sandy Beachcomber: if i still get viewer crashes i will go and eat worms
[12:21] Bill Blight: need to put the SL mainland mentality to bed .. Slap a 8x8 var down and call it mainland
[12:22] Ubit Umarov: 10x is a bit 2 large
[12:22] Bill Blight: 10x10 works but viewer hates it ... LOL
[12:22] Leighton.Marjoram working example not a real suggestion
[12:22] Ubit Umarov: do not forget that on very large regions you may get issues due to precision
[12:22] Arielle Popstar: how do you know?
[12:22] Ubit Umarov: similar to the ones you get at SL editing at very high Z
[12:22] Bill Blight: I have done regions that big and bigger for testing
[12:22] Sandy Beachcomber: its difficult to terraform large vars
[12:23] Arielle Popstar: me too and the viewer showed it and didnt crash
[12:23] Andrew Hellershanks: I've also seen the occasional 10x10 var.
[12:23] Ubit Umarov: on lar regions you will get those issues on X and Y
[12:23] Arielle Popstar: so how do you know it hates it
[12:23] Kayaker Magic: The terrain in large vars loads slow, need a lot of bandwidth and memory.
[12:23] Ubit Umarov: ( worse as X and Y is higher )
[12:23] Bill Blight: I have 9 16x16 vars I put together just for testing , viewer lost it's mind
[12:24] Andrew Hellershanks: That's a lot of real estate.
[12:24] Bill Blight: yeppers
[12:24] Kayaker Magic: Do the math: How much memory to store width*height floating point z values, plus the size of the texture map.
[12:24] Ubit Umarov: guess 2km is the real max
[12:24] Ubit Umarov: 1km possible the best tradeoff
[12:24] Arielle Popstar: till we get bigger servers
[12:25] Kayaker Magic: You also need more memory for the viewer!
[12:25] Bill Blight: 24cores and 256 gig of ram not big enough for you?
[12:25] Ubit Umarov: ( the issue im talking about has nothing to do with size of servers, terrain size maybe )
[12:25] Gavin.Hird try the new Mac Pro with 1.5 TB memory
[12:26] Bill Blight: yes, even on an empty region the viewer wanted and did many times run out of ram and crash or slowed to a crawl
[12:27] Andrew Hellershanks: We are at about half past now. Time to call this meeting to a close.
[12:27] Kayaker Magic: Great, build a grid with large vars and only people who buy brand new computers can log onto it.
[12:27] Andrew Hellershanks: Thank you all for coming. See you again next week.]

Personal tools
About This Wiki