Chat log from the meeting on 2017-01-24

From OpenSimulator

Revision as of 13:30, 24 January 2017 by Sheera Khan (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

[11:06] Andrew.Hellershanks @hg.osgrid.org: Finally found the chair with an area select. :)
[11:06] George Equus: Selected a seat at far end, ended up here...
[11:07] Andrew.Hellershanks @hg.osgrid.org: I'm going to relog. Singularity is using a lot of CPU.
[11:08] Misterblue Waves: does this region really have a day cycle of "None"?
[11:08] Marcus Llewellyn: It shouldn't.
[11:08] Marcus Llewellyn: Or didn't used to.
[11:08] James.atLLOUD @hg.osgrid.org: I saw the same thing @ day cycle. funny
[11:08] Sheera Khan: I didn't ever experience a night here
[11:09] Misterblue Waves: I've often found that Firestorm/Windlight/OpenSimulator don't play well together
[11:09] Marcus Llewellyn: You can always override if it takes yer fancy. :)
[11:09] Kayaker.Magic @grid.kitely.com:8002: Overriding the sun position always resets every time you cross a sim border.
[11:10] Marcus Llewellyn: The horror. ;)
[11:10] Kayaker.Magic @grid.kitely.com:8002: any way to make it stick?
[11:10] Kayaker.Magic @grid.kitely.com:8002: It is annoying to be sailing a boat along and every 256 meters the sun goes down and you have to stop to bring it back up.
[11:10] Mike.Lorrey @grid.kitely.com:8002: good thing you don't have to navigate by it, eh?
[11:11] Kayaker.Magic @grid.kitely.com:8002: Another good reason for scripting a good in-world GPS
[11:11] James.atLLOUD @hg.osgrid.org: I've grown terrible dependent on Firestorm Phoenix button - all environmentals at a time.
[11:12] Marcus Llewellyn: I've been using FS lately because it's been most stable for me. But it's not what I would prefer.
[11:12] Kayaker.Magic @grid.kitely.com:8002: I use FS for the LSL pre-processor.
[11:12] Marcus Llewellyn: It is awful nice for scripting. And it's renderer is prolly the best of the bunch.
[11:12] James.atLLOUD @hg.osgrid.org: I use firestorm mostly out of habit. lol
[11:13] Andrew.Hellershanks @hg.osgrid.org: I've been using Singularity mostly out of habit.
[11:13] Mike.Lorrey @grid.kitely.com:8002: the new bento release seems to run a lot smoother with less memory use
[11:13] Kayaker.Magic @grid.kitely.com:8002: nice to hear
[11:13] Marcus Llewellyn: Alchemy is top of my preference list, but it's been a bit on the crashy side. :(
[11:14] Andrew.Hellershanks @hg.osgrid.org: I've been doing some building in the last few days. I should try using FS 5 to give it a try.
[11:14] Andrew.Hellershanks @hg.osgrid.org: Problem with switching viewers is finding out where the viewer devs moved some the features. Different viewers, different layout for menu and preferences/
[11:15] James.atLLOUD @hg.osgrid.org: FS 5 at home and at work with very little problems.
[11:15] Mike.Lorrey @grid.kitely.com:8002: agreed, I used Singularity the other day to test if a bug was viewer side or server side, and took a few minutes to find where things were
[11:15] Marcus Llewellyn: FS is the "and the kitchen sink" of viewers... prolly the worst in terms of settings hunting. So much so that they added a search function to preferences.
[11:15] Kayaker.Magic @grid.kitely.com:8002: Andrew, did anyone answer my question from last week about putting var regions of different size next to each other?
[11:15] Selby.Evans @grid.kitely.com:8002: FS 5 works well for me
[11:15] James.atLLOUD @hg.osgrid.org: Like photoshop! lols
[11:16] Kayaker.Magic @grid.kitely.com:8002: Photoshop is a verb, it is what I to do images with GIMP.
[11:16] Mike.Lorrey @grid.kitely.com:8002: are we waiting to start or am I missing things on voice?
[11:16] Kayaker.Magic @grid.kitely.com:8002: never in voice
[11:16] James.atLLOUD @hg.osgrid.org: agreed. good also to photoshop alternative facts.
[11:16] Andrew.Hellershanks @hg.osgrid.org: Kayaker, yes. I'm trying to remember the response. :)
[11:17] Marcus Llewellyn: I try to use GIMP, I really do. But I can't. It just kills me. 2.9.x looks like it'll be a huge improvement, when it ever gets out the damn door.
[11:17] Kayaker.Magic @grid.kitely.com:8002: I almost could not remember my question!
[11:17] Andrew.Hellershanks @hg.osgrid.org: Kayaker, it might work but there are likely to be problems with the viewer.
[11:17] Kayaker.Magic @grid.kitely.com:8002: Too much going on in my lives.
[11:17] Asdf Jkl: try one window mode in gimp
[11:17] James.atLLOUD @hg.osgrid.org: Kayaker, I had the impression you'd tried the VARS -
[11:17] Andrew.Hellershanks @hg.osgrid.org: Marcus, for a moment I thought you were about to say 2.9 was going to be worse in some way.
[11:17] Marcus Llewellyn: One window mode doesn't fix the other headaches. It shouldn't take 12 steps just to stroke some text.
[11:18] Mike.Lorrey @grid.kitely.com:8002: Now that I've learned to use Blender enough that I no longer feel that it was designed for autistics, I might give learning GIMP a try
[11:18] Andrew.Hellershanks @hg.osgrid.org: I've used GIMP in multi-window mode since I first ran across it. I don't like single window mode. I find it too restrictive.
[11:18] Andrew.Hellershanks @hg.osgrid.org grins at Mike.
[11:18] James.atLLOUD @hg.osgrid.org wants multi window OS viewer.
[11:18] Marcus Llewellyn: Amen James.
[11:18] Andrew.Hellershanks @hg.osgrid.org: Good on you for getting a handle on Blender.
[11:19] Kayaker.Magic @grid.kitely.com:8002: I was visiting a grid where the owner had set 4x4 vars next to 8x8 and sometimes it worked. I wondered if this was recommended, if it was going to be supported better in the future.
[11:19] Kayaker.Magic @grid.kitely.com:8002: Should we submit mantis reports about it?
[11:20] Marcus Llewellyn: Var is such a mix of server and viewer... that makes crossing issues difficult to track down.
[11:21] Andrew.Hellershanks @hg.osgrid.org: Kayaker, you could mantis it. From what I recall it is more a viewer issue than an OS one.
[11:21] Alicia.Raven @grid.spellscape.co.uk: different var sizes close together is a viewer problem i think, the viewer takes the region size and expects the child regions to be the same size
[11:21] Misterblue Waves: var regions of different sizes will fit together and work
[11:21] Misterblue Waves: there are known problems, though, mostly with visibiltiy, view distance,...
[11:22] Andrew.Hellershanks @hg.osgrid.org: I know some people like to have a very large draw distance. I keep mine around 64m or so.
[11:22] Misterblue Waves: that is, one can start in a larger region and see smaller regions but, once borders are crossed, you might not be able to see some of the other regions
[11:22] Kayaker.Magic @grid.kitely.com:8002: OK, I'll mash a few together and see if I can generate some Mantises about it.
[11:22] James.atLLOUD @hg.osgrid.org: 128 m draw dist today
[11:22] Marcus Llewellyn: Viewers were just never designed to see more than a couple 256 regions out. Make a var larger than 1024 and you run into things like skydome clipping.
[11:23] Misterblue Waves: I suspect the problem is with the computation of child regions as it related to a users set draw distance, the servers 'default view distance' and comm knowing the sizes of regions around a simulator
[11:25] James.atLLOUD @hg.osgrid.org: child regions being whatever regions are ouside the current region?
[11:26] Kayaker.Magic @grid.kitely.com:8002: Well, that is the only question I have time for today, I have some RL appointments to get to.
[11:26] Simulator Version v0.5 ruft: OpenSim 0.9.1.0 Dev 8d99c86: 2017-01-22 05:07:43 +0000 (Unix/Mono)
[11:26] Andrew.Hellershanks @hg.osgrid.org: James, they are the neighbouring regions beside the region you are in.
[11:26] Misterblue Waves: yes... the region your avatar is on is the 'root region'.... the viewer is told about surrounding 'child region's so you can see what's over there
[11:27] James.atLLOUD @hg.osgrid.org: So I saw a bunch of code commits from Ubit about permissions.
[11:28] Andrew.Hellershanks @hg.osgrid.org: Yes. There were quite a few more related to permissions this past week.
[11:28] Andrew.Hellershanks @hg.osgrid.org: The permissions code is getting a fairly thorough going over.
[11:29] James.atLLOUD @hg.osgrid.org: yeah, that's impressive.
[11:29] Andrew.Hellershanks @hg.osgrid.org: It is long overdue.
[11:29] Asdf Jkl: how do you compile ubODE?
[11:29] Misterblue Waves: Mel and Ubit are making sure permissions are correct before the next release... it's quite an effort
[11:30] Asdf Jkl: did not find source for ubODE
[11:30] Andrew.Hellershanks @hg.osgrid.org: Asdfjkl, ubODE is part of the core of Open Simulator. It gets compiled automatically when you build OpenSim. You just need to change a couple of settings in the OpenSim.ini file if you want to use that physics engine instead of the default BulletSim engine.
[11:30] Misterblue Waves: the C# portion of ubODE is in the OpenSimulator sources... you asking about the C++ portion?
[11:31] Asdf Jkl: about the lib
[11:31] Asdf Jkl: .so files
[11:32] Andrew.Hellershanks @hg.osgrid.org: Let me see if Ubit is available
[11:32] Sheera Khan: apropos, what happened to the project of changing the location?
[11:32] Marcus Llewellyn: It might be in opensim-libs
[11:33] Sheera Khan: ;-)
[11:33] Misterblue Waves: ubODE uses the regular ODE binaries.... those are distributed in bin/lib32 and bin/lib64
[11:33] Asdf Jkl: ok so you compile ode
[11:34] Misterblue Waves: the sources for the ODE we use are in the library repository: git://opensimulator.org/git/opensim-libs
[11:34] Asdf Jkl: so i use ubODE and not ODE in opensim.ini ?
[11:35] Andrew.Hellershanks @hg.osgrid.org: I'm not finding ubODE in either libOMV or in opensim-libs
[11:35] Misterblue Waves: in 'opensim-libs' are the sources for the compiled DLLs and SOs... httpserver, OpenDynamicsEngine, BulletSim, OpenJPEG, ...
[11:35] Marcus Llewellyn: http://opensimulator.org/viewgit/?a=tree&p=opensim-libs&h=3aebc0d6be17798c6a23000b28e415a1b6b8f7b2&hb=017c52aec138c80820e8683daab82f38ccc6ac04&f=trunk/unmanaged
[11:36] Misterblue Waves: The physics engines have two halves: the C# source that is with the simulator code, and a C++ library that is distributed as a binary
[11:37] Marcus Llewellyn: Most people never have to pull from opensim-libs unless they're building for a different architecture, or something like FreeBSD. The regular opensim git repo just has binaries for unmanaged code.
[11:37] Misterblue Waves: The C# sources for ubODE are with the simulator code in OpenSim/Region/PhysicsModules/ubOde/
[11:38] Asdf Jkl: so if i want to use ubODE with 0.9 i simply use ODE ?
[11:38] Andrew.Hellershanks @hg.osgrid.org: ODE is the old physics engine that predates BulletSim
[11:38] Misterblue Waves: the C++ sources for the ODE physics engine (that is distributed in bin/lib32 and bin/lib64) are in opensim-libs/trunk/unmanaged/OpenDynamicsEngine-0.13.1mod
[11:39] Andrew.Hellershanks @hg.osgrid.org: Asdfjkl, Look at the comments in OpenSim.ini in [Startup] just above the line "physics ="
[11:39] Misterblue Waves: there are three physics engines availabe in OpenSimulator: ODE, ubODE, and BulletSim
[11:39] Asdf Jkl: ok. thx.
[11:39] Marcus Llewellyn: Did we finally throw POS away completely?
[11:40] Misterblue Waves: the first two use the same binary physics engine but ubODE has been tuned to better simulate SL physics
[11:40] Andrew.Hellershanks @hg.osgrid.org: Mister, oh it just has a dirname ending in mod? I was looking for a directory with "ub" in the name.
[11:40] Misterblue Waves: well, there are the old physics engines: 'no physics' and 'basic physics'... but those don't count :)
[11:41] Marcus Llewellyn: Hehe
[11:41] Andrew.Hellershanks @hg.osgrid.org: That gets confusing. There are three ODE directories in the OS libs tree
[11:41] Marcus Llewellyn: I remember those rather fondly. ;)
[11:41] Misterblue Waves: ODE => OpenDynamicsEngine
[11:42] Andrew.Hellershanks @hg.osgrid.org: The OpenSim.ini also mentions Bullet and BulletX physics engines, IIRC
[11:43] Misterblue Waves: OpenSimulator sources don't throw anything away.... it has become cluttered over the years
[11:43] Marcus Llewellyn: Basically, unless you're patching it, building it for a different architecture (say ARM) or a different *NIX, you shouldn't ever need to compile those binaries yourself.
[11:43] Andrew.Hellershanks @hg.osgrid.org: Asdfjkl, if you are wanting to help with development of the ubODE physics engine you should join the opensim-dev channel on IRC and talk with Ubit.
[11:44] Misterblue Waves: For me, finding a system to compile the Apple versions is hard :)
[11:45] Asdf Jkl: ok
[11:46] Andrew.Hellershanks @hg.osgrid.org: Misterblue, no one has sent you an Apple based development system yet? ;)
[11:47] Marcus Llewellyn: People can lend me machines. I could use a few Xeon servers. :D
[11:48] James.atLLOUD @hg.osgrid.org: lol -those are just laying around.
[11:48] Asdf Jkl: so 0.13 is ubODE and 0.10 is ODE?
[11:48] Misterblue Waves: maybe I can find an old Apple machine for sale somewhere... or see if one of the cloud services has IOS VMs
[11:49] Misterblue Waves: Askf: I'm not sure.... Ubit would know
[11:49] Andrew Hellershanks: and the -r1755 is something else
[11:49] Marcus Llewellyn: Yeah.... there's always the hackintosh (physical or VM) approach. As long as licensing doesn't bother you.
[11:49] Asdf Jkl: or is both ODE and ubODE is only the bridge in opensim
[11:49] Asdf Jkl: ok
[11:52] Andrew Hellershanks: In the ODE r1755 directory is a file called WrongVersion.txt that seems a little confusing as to whether it or the 0.10.1 is more up to date.
[11:53] Andrew Hellershanks: The r1755 may be 0.12 of ODE physics
[11:53] Marcus Llewellyn: I wouldn't be surprised if there's isn't more than a few inconsistant things in opensim-libs. The only people who ever venture there are core.
[11:53] Andrew Hellershanks nods
[11:54] James atLLOUD: beyond there lay dragons
[11:54] Andrew Hellershanks: The last time I really went in to the opensim-libs source tree was in March of 2014.
[11:54] Misterblue Waves: there's a new version of the Bullet physics engine coming out and I'll probably update BulletSim then.... I have a few things to fix so that will make me take a look at it
[11:55] Marcus Llewellyn: Although if anyone wants to use Nini in a project, I'd recommend using opensim's. That lib has long been abandoned, and Bluewall made quite a few fixes.
[11:56] Andrew Hellershanks: I don't know what features are included in Nini besides parsing an INI file. I wrote my own ini parser when I was working on a project.
[11:56] Andrew Hellershanks: One of the features of Nini I consider annoying and almost class as a bug.
[11:57] Marcus Llewellyn: Yeah, and INI parser is pretty easy, and I wrote one that can handle Opensim's INI "helper" lines. But it can also do XML or read from http, which is nice.
[11:58] Andrew Hellershanks nods
[11:58] Misterblue Waves: we started pulling sources for libraries OpenSimulator was dependent on into opensim-libs because they do get abandoned after many years -- OpenSimulator has been around and active for a long time, open source wise
[11:59] Andrew Hellershanks: Sure. So now code can be mostly abandoned in opensim-libs. ;)
[11:59] Marcus Llewellyn: lol
[11:59] Andrew Hellershanks: It is good to see occasional commits in it. It doesn't see many updates.
[12:01] Andrew Hellershanks: I plan to do some work on the LSL parser one of these days. That will involve some digging around in opensim-libs.
[12:01] Misterblue Waves: most of the work I've seen in opensim-libs has been on fixing the HTTP server.
[12:01] Marcus Llewellyn: Start from scratch. :P Mel did a great job replacing our original scripting engine, but the one we have is still so damn quirky.
[12:02] Marcus Llewellyn: Make a sandboxed and secure c# scripting engine, and I'll have your babies.
[12:03] Andrew Hellershanks: I have a cat. I don't need any babies. :)
[12:03] Marcus Llewellyn: lol
[12:03] Misterblue Waves: I've wondered about porting the Halcyon script engine over.... there is also the old one from Aurora that would be cool, too.
[12:03] James atLLOUD: (!)
[12:03] Marcus Llewellyn: Aurora's was kinda.... adhoc, although it did have some nice features (arrays!). Halcyon might be better.
[12:04] Andrew Hellershanks: I'm mostly interested in some of the parsing or LSL to C# translation errors.
[12:04] Andrew Hellershanks: There are some issues with how vector math is handled.
[12:04] Misterblue Waves: one of the best features of the existing script engine is its extensibilitiy -- a region module can easily add functions to LSL
[12:04] James atLLOUD: So it seems to me this discussion revealed an amount of un-needed code?
[12:05] Misterblue Waves: which code is un-needed, James?
[12:05] Marcus Llewellyn: Not unneded per say, Legacy in some cases, or transitory in others.
[12:05] James atLLOUD: OK, thanks Marcus.
[12:06] Misterblue Waves: since people extend OpenSimulator, there are always problems with changing APIs or removing code because someone out there might depend on it
[12:06] Misterblue Waves: there have been recent complaints about money modules breaking because some APIs inside OpenSimulator changed
[12:06] James atLLOUD: sure, I understand.
[12:07] Misterblue Waves: so we tend to error on the side of letting sleeping dogs lie
[12:07] James atLLOUD: got it.
[12:07] Andrew Hellershanks: Someone on IRC recently asked about the DTL/NSL money module with 0.9.1
[12:07] Misterblue Waves: that's why there are three or four statistics gathering systems in the sources as well as a few HTTP access entry points and on and on
[12:08] Andrew Hellershanks: The changes related to modules and/or the sample money system in OS are minor. It doesn't take a lot of work to update most modules.
[12:09] Andrew Hellershanks: The common problem is some declarations near the start of a module.
[12:09] Asdf Jkl: so i need also another scripting engine for 0.9 to be able to use vehicle Scripts?
[12:09] Andrew Hellershanks: I had to make a minor update to the OpenSimSearch and OpenSimProfile modules at one time.
[12:10] Andrew Hellershanks: You don't need a different scripting engine.
[12:10] Marcus Llewellyn: No, there's just the one engine. But physics functions behave differently, or are have more/less support depending on which engine you choose to use.
[12:10] Misterblue Waves: ODE, ubODE and BulletSim all have the SL vehicle commands implemented
[12:10] Andrew Hellershanks: Vehicle scripts are more dependant on the physics engine, if the scripts uses physics.
[12:10] Asdf Jkl: ok
[12:11] Andrew Hellershanks: Not all vehicle scripts use physics.
[12:12] Misterblue Waves: OpenSimulator can have multiple script engines -- a feature few people use
[12:12] Andrew Hellershanks: I just noticed we are now past the hour mark. Are there any more topics to raise for todays meeting?
[12:13] Misterblue Waves: there have been previous, incomplete efforts for python, C#, and others
[12:13] Marcus Llewellyn: I've pondered a node.js bridge from time to time.
[12:14] Misterblue Waves: it would be 'easy' to have a region module that added new LSL functions and provided external communications -- the XEngine is easily extensible that way
[12:15] Marcus Llewellyn: Quiet a few advantages to the idea. Offloading scripts to a seperate machine would be nice. node.js also allows for easy white/blacklisting.
[12:15] Andrew Hellershanks: IIRC, you can write scripts in some other languages but that opens up another can of worms. It is not a feature enabled by default.
[12:16] Marcus Llewellyn: Yeah, it's the lack of whitelisting that kills using C#.
[12:16] Misterblue Waves: I'd looked at using Thrift to export some of the internal APIs... for, like, moving physics off to another machine
[12:16] Marcus Llewellyn: Not familiar with Thrift
[12:16] Andrew Hellershanks: That's a new one to me.
[12:16] Misterblue Waves: enhancements not being everywhere is the major impediment to adding new things
[12:17] Marcus Llewellyn: Apache project?
[12:18] Misterblue Waves: ya... Thrift is one of those compiling remote API systems -- you define the API then compile the interface for server and clients
[12:18] Misterblue Waves: Thrift is active and has targets for C++, C#, JavaScript, Python, Java, ...
[12:19] Marcus Llewellyn: INteresting!
[12:19] Marcus Llewellyn: But yeah, we're into pie in the sky stuff now. Meeting should end. lol
[12:20] Andrew Hellershanks: ok, on that note I'll call this meeting concluded for this week.

Personal tools
General
About This Wiki