Chat log from the meeting on 2015-03-10

From OpenSimulator

Jump to: navigation, search
[11:02]  Marcus Llewellyn: I don't see any errors. I'm not HGed in though.
[11:02]  Shez Oyen: I'm am getting a flood of .. "Primitive: llSetPrimitiveParams: Error running rule #2: arg #0 - unsupported parameter"
[11:02]  Connecting to in-world Voice Chat...
[11:02]  Connected
[11:02]  Hiro Protagonist is Online
[11:02]  Shez Oyen: someones attatchments are wonkie?
[11:02]  OtakuMegane Desu: I'm getting those too
[11:02]  Andrew Hellershanks: Someone must have messed up the rules in the call, Shez
[11:02]  Connecting to in-world Voice Chat...
[11:02]  Connected
[11:02]  Aine.Caoimhe no, it's been happening for the last few weeks
[11:02]  Shez Oyen: wow
[11:02]  Aine.Caoimhe any time anyone enters the region it throws up a new one
[11:03]  Aine.Caoimhe only happens here...not on any other grids
[11:03]  OtakuMegane Desu: Anyone else encountered border crossing issues when offline regions are nearby?
[11:03]  Aine.Caoimhe but it happens in all OSG regions
[11:03]  Shez Oyen: Yes Otaku!!
[11:03]  Shez Oyen: I have been griping about that for months
[11:04]  Justin Clark-Casey: hello folks
[11:04]  Marcus Llewellyn: Heya Justin
[11:04]  Shez Oyen: that many people aren't entering :/
[11:04]  BlueWall Slade: Hi jcc
[11:04]  Hiro Protagonist: hey Justin :D
[11:04]  Shez Oyen: Hi Justin :)
[11:04]  OtakuMegane Desu: Teravus and Seaprior Plazas are a good demo at the moment
[11:05]  OtakuMegane Desu: I did track down the commit it happened in. There was a change that delayed agent transfer until all the communication was done because otherwise some things could get messed up.
[11:05]  OtakuMegane Desu: Unfortunately that apparently means if child agents in nearby regions don't respond, you can't go anywhere
[11:06]  Shez Oyen: I wish "someone" would fix that Otaku! a lot of my neighbrs crash on purpose to save their spots.. makes region line crossing a real pain on your own regions
[11:07]  OtakuMegane Desu: Well with OSGrid down soon after that commit, offline regions are not probably a big thing most other places so it doesn't show up so much.
[11:07]  Andrew Hellershanks: Oh, I see those errors in the debug dialog box. Someone should check the region console and find out what device is throwingthe errors.
[11:07]  OtakuMegane Desu: People are probably running into it a lot now but chalk it up to the usual bugginess or something.
[11:07]  Justin Clark-Casey: otaku: which commit?
[11:07]  Aine.Caoimhe it's not a's a profile fetch error on every single agent entering the region
[11:07]  OtakuMegane Desu: Gimme a sec
[11:08]  Ezequiel Menna: oh Aine
[11:08]  Ezequiel Menna: are you Danger friend?
[11:08]  Aine.Caoimhe yes
[11:08]  Ezequiel Menna: i see :)
[11:08]  AlexRime SiLiSiLi is Online
[11:08]  OtakuMegane Desu: 4c5d7d4683e34f9ac061da649ebf29e23d612faf
[11:08]  Andrew Hellershanks: The messages about errors in llSetPrimitiveParams should go to the owner of the device and not to everyone in the region
[11:08]  OtakuMegane Desu: Back on 7/10 last year
[11:09]  Andrew Hellershanks: Ah, the errors are coming from a device near 52.45, 213.19
[11:09]  Justin Clark-Casey: otaku: I see. Do you know if there is a Mantis for this?
[11:09]  Andrew Hellershanks: Too bad it didn't show the Z
[11:10]  OtakuMegane Desu: In console it's usually Slow-RPC messages that take 10-30 seconds to timeout that I rarely see any other time
[11:10]  OtakuMegane Desu: And no, I haven't found any other mantis about it
[11:10]  Shez Oyen: Robert I hope you mange to rez.. it's almost good to see you ;)
[11:10]  OtakuMegane Desu: Haven't had a chance to file one myself XD
[11:10]  Justin Clark-Casey: otaku: so there is no bug report for it?
[11:11]  Nebadon Izumi is Online
[11:11]  Aine.Caoimhe I was kind of curious to find out how long the inventory fetch would continue on my most recent hang....I left it overnight and it was still going strong when I got up the morning so if you'd like a nice 46GB log file I can arrange for a truck to ship it to you
[11:11]  Nebadon Izumi: hello
[11:11]  Justin Clark-Casey: otaku: ok. It woudl be good to have one.
[11:11]  Nebadon Izumi: sorry I am late
[11:11]  OtakuMegane Desu: Nope, not to my knowledge
[11:11]  Justin Clark-Casey: hello nebadon
[11:11]  BlueWall Slade: Hi Neb
[11:11]  OtakuMegane Desu: Yeah
[11:12]  OtakuMegane Desu: I was trying to poke at it a bit more before making one and got sidetracked by life for a while
[11:12]  Shez Oyen: Greetings Neb :)
[11:12]  AlexRime SiLiSiLi: Hello all:)
[11:12]  BlueWall Slade: Hi AlexRime
[11:12]  Shez Oyen: Hi Alex :)
[11:13]  Justin Clark-Casey: otaku: yeah thanks. Good concise info is very helpful, though identifying the commit is very good. I can imagine offline regions may be an issue with this change
[11:13]  Justin Clark-Casey: otaku:so what happens - the agent just slides against the border?
[11:14]  OtakuMegane Desu: Yeah. It behaves kind of like when there's no region at all. Once the requests time out then you can cross. But going back you have to wait all over again.
[11:14]  Justin Clark-Casey: and this happens on the first try?
[11:14]  Shez Oyen: I tried shortening that time-out but it didn't help enough to notice
[11:14]  OtakuMegane Desu: Yeah. And most every subsequent try as well.
[11:15]  Justin Clark-Casey: otaku: and this is with neighbours that are registered but not actually online?
[11:15]  Shez Oyen: Yes Justin
[11:16]  OtakuMegane Desu: Yeah. Seaprior and Teravus plazas both have at least one or two neighbors down since the grid came back up and have been a pretty good demo so far
[11:17]  Justin Clark-Casey: otaku: ok, I will try and take a look. I made a not disimilar change recently (beef41f) which stopped sims trying to contact registered but offline regions in another context. Possibly the same could be done here
[11:17]  Justin Clark-Casey: but it would be very good to have a bug report to track this as well
[11:18]  OtakuMegane Desu: After the meeting I'll see if I can dig up the console messages and make one.
[11:18]  Simulator Version v0.5 shouts: OpenSim Dev        00b5b91: 2015-03-07 17:47:40 -0800 (Unix/Mono)
[11:18]  Shez Oyen: Robert fixed it so an unwary new person could not accidently set up their regular region so that it overlaps a varregion. Does this not work the other way around? Like can a new var overlap another persons regions? Seems logical that it would work both ways but does it?
[11:19]  Justin Clark-Casey: it should work that way too
[11:20]  Justin Clark-Casey: z
[11:21]  Robert Adams: I believe it does check for any overlap
[11:21]  Shez Oyen: these errors are like every second?
[11:21]  Shez Oyen: Thanks Robert.. there are people griping on the forum that vars are overlapping them
[11:21]  Shez Oyen: I didn't think that was possible
[11:22]  Justin Clark-Casey: if i remember there is some config that can turn the check off... but I thought the default was to have the check
[11:23]  Freaky Tech: hi
[11:23]  Shez Oyen: Hiya Freaky :)
[11:23]  BlueWall Slade: couldit just be child agents in their regions fromthe var?
[11:23]  BlueWall Slade: Hi Freaky Tech
[11:23]  Justin Clark-Casey: hello freaky
[11:23]  Freaky Tech: so unused to that slow loading of inventory ;)
[11:25]  Freaky Tech: neever mind just takes longer to bake then ;)
[11:27]  Justin Clark-Casey: z
[11:27]  Nebadon Izumi: anyone who is interested and did not see my email on the mailing lists earlier, the OSCC grid is fully operational again
[11:28]  Andrew Hellershanks: oh, right. it was about the OSCC grid. That didn't quite register.
[11:28]  AlexRime SiLiSiLi: Than was the reason we has no server backup?
[11:28]  Andrew Hellershanks: Sort of thought it was still OSG related at first.
[11:28]  AlexRime SiLiSiLi: I cost us half year of offline
[11:29]  Marcus Llewellyn: OSCC grid, AlexRime. Not OSGrid.
[11:29]  Nebadon Izumi: yea its a different grid
[11:29]  Nebadon Izumi: the OpenSImulator Community Conference grid
[11:29]  Nebadon Izumi:
[11:29]  AlexRime SiLiSiLi: And OSCC have server backup now:)?
[11:29]  Nebadon Izumi: to hypergrid
[11:29]  Marcus Llewellyn: OSCC really only needs to be up for the conference. Uptime other than that is just a bonus. :)
[11:29]  Aine.Caoimhe it appears that functionality of setting a region to disallow HG direct tp is broken
[11:30]  Nebadon Izumi: ya it was down intentionally
[11:30]  Nebadon Izumi: for upgrades
[11:30]  Nebadon Izumi: had to reload Linux
[11:30]  BlueWall Slade: what dod you load on it?
[11:30]  Nebadon Izumi: opensuse 13.1
[11:30]  Andrew Hellershanks: Data recovery worked well. I don't have a lot of inventory but I seem to have all the items I had before.
[11:30]  BlueWall Slade: +
[11:30]  Marcus Llewellyn: Raspian! >:)
[11:30]  Nebadon Izumi: it was running 11.4
[11:30]  BlueWall Slade: wow
[11:30]  Nebadon Izumi: had kernel 2.6 lol
[11:30]  Aine.Caoimhe trouble is I haven't had time to do several hundred builds to figure out what git it was broken in
[11:30]  Nebadon Izumi: tad bit old
[11:31]  Andrew Hellershanks: Marcus, Raspian. :D
[11:31]  Nebadon Izumi: has mono 3.12.1 now and the latest version of opensim
[11:31]  Marcus Llewellyn: Next year it should be hosted all on Pis, says I. ;)
[11:31]  Nebadon Izumi: i did move my Raspberry Pi region here to OSgrid now
[11:31]  Marcus Llewellyn: Neato. :)
[11:31]  Freaky Tech: did anyone of you take a peek on ?
[11:32]  Nebadon Izumi: OKC Raspberry Pi2
[11:32]  Andrew Hellershanks: I have my Pi B but I'm still looking at getting a 2B
[11:32]  Nebadon Izumi: I saw that Freaky i had planned to take a look at it myself
[11:32]  Nebadon Izumi: but have not had time yet
[11:32]  Nebadon Izumi: maybe someone will get a chance to test it before i can
[11:33]  Freaky Tech: test it on var regions as well because it tracks what is sent to viewer to the agent about what terrain patch
[11:33]  Freaky Tech: and tracks changes
[11:33]  Nebadon Izumi: nice
[11:33]  Freaky Tech: despite fixing a MTU bug inside terrain patch sending
[11:35]  BlueWall Slade: that one slipped by me :)
[11:35]  Billy.Bradshaw We have the patch running on several regions, both var and regular sizes. We have not specifically tested for the issues it addresses, but seems to run without issue.
[11:35]  Nebadon Izumi: i think it got added yesterday?
[11:35]  Nebadon Izumi: to mantis that is
[11:35]  Nebadon Izumi: that was when i first noticed anyway
[11:35]  Freaky Tech: yes was yesterday
[11:35]  BlueWall Slade: +
[11:35]  BlueWall Slade: I know we talked about it.
[11:35]  Freaky Tech: bug fixing and implementation was around saturday
[11:36]  Nebadon Izumi: cool Billy
[11:36]  BlueWall Slade: I'll test it after the meeting
[11:36]  Freaky Tech: porting was yesterday
[11:36]  Nebadon Izumi: cool thanks BlueWall
[11:36]  Nebadon Izumi: ive been very busy
[11:36]  Freaky Tech: a standard region with flat terrain loads in three UDP messages with that patch
[11:36]  BlueWall Slade: me too :s
[11:36]  Nebadon Izumi: still wrapping up from FCVW too
[11:36]  Freaky Tech: whereas the old code needed 64
[11:36]  BlueWall Slade: ohh yeah
[11:37]  Justin Clark-Casey: robert: why is OpenSimTerrainCompressor OpenSim/Region/Framework/Scenes but in the OpenSIm.Region.ClientStack.LindenUDP namespace?
[11:37]  Billy.Bradshaw yes we did observe the reduction in data
[11:37]  Freaky Tech: Justin that patch moves the TerrainCompressor anyways
[11:37]  Justin Clark-Casey: ft: yes, though it would seem to me very lindenudp related, so I'm wondering why it isn't there
[11:38]  Freaky Tech: the problem is HeightmapTerrainData is not there so TerrainCompressor has mainly to go where the compressed terrain patches are hold
[11:38]  Freaky Tech: for continuous usage
[11:38]  Hiro Protagonist: I'm sorry but I cant get a browser to play nice and open that link
[11:39]  Hiro Protagonist: does that have to do with the problem of sending large blocks of region info and vars blocking after right clicks in the scene?
[11:40]  Freaky Tech: it tracks changes and what an avatar has already got of the terrain data
[11:40]  Freaky Tech: so no resend when it is already transmitted
[11:40]  Justin Clark-Casey: ft: where is that forced? I don't see any obvious reason why it couldn't be in LindenUDP
[11:40]  Freaky Tech: your assembly structure
[11:40]  Freaky Tech: I cannot link OpenSim.Framework to Opensim.Region.Framework and back
[11:40]  Freaky Tech: circular dependency issue
[11:41]  Justin Clark-Casey: where do you need to do that?
[11:41]  Freaky Tech: to keep the terrain patches hold for longer use
[11:41]  Freaky Tech: I have to keep track on the HeightmapTerrainData what has changed
[11:41]  Nebadon Izumi: for terrain editing?
[11:41]  Freaky Tech: and compress only when needed
[11:42]  Freaky Tech: HeightmapTerrainData knows about updates
[11:42]  Freaky Tech: adding another event mess would be just horrid
[11:42]  Justin Clark-Casey: ft: ok, I see you make a call in TerrainData.GetCompressedPatch()
[11:42]  Freaky Tech: because it is hard to track
[11:42]  Freaky Tech: exactly
[11:42]  Justin Clark-Casey: yeah, but adding UDP stuff into OpenSim.Framework is also a long term pita
[11:42]  Justin Clark-Casey: maybe fine in the short term but in the best world would not be that way forever
[11:43]  Freaky Tech: Justin when I look through the whole code the Linden specific stuff is everywhere
[11:43]  Dahlia Trimble is Online
[11:43]  Justin Clark-Casey: yes, though there's a differene between linden stuff and linden udp stuff
[11:43]  Freaky Tech: Justin does it make sense to kill lots of CPU power by recompressing terrain on every terrain sending?
[11:43]  Justin Clark-Casey: I'm not saying there might not also be udp stuff in OpenSim.Framework but that in itself is an eyesore
[11:44]  Freaky Tech: to get that stuff out there it would have required a far larger patch
[11:44]  Dahlia Trimble: hi
[11:44]  Justin Clark-Casey: ft: no, I'm not saying that. But I'm sure tehre's a better solution that doesn't involve stuffing things in Framework
[11:44]  Justin Clark-Casey: ft: yes, and that should be separate - but at some point such a patch should exist
[11:44]  Shez Oyen: Dahlia :)
[11:44]  Freaky Tech: actually HeightmapTerrainData says that the actual code is Linden optimized
[11:45]  Freaky Tech: with that extreme separation you actually added borders for some pieces being able to work well
[11:46]  Justin Clark-Casey: me?
[11:46]  Freaky Tech: and you know what is a major eyesore and headache
[11:46]  Freaky Tech: calling namespaces Framework
[11:46]  Freaky Tech: they tend to get filled like a mixed supermarket
[11:47]  Freaky Tech: so just to give an oppinion what I think about namespacing. do not feel offended by that.
[11:47]  Dahlia Trimble: lots of eyesores in opensimulator code lol
[11:47]  Justin Clark-Casey: ft: I don't, though I don't really agree
[11:47]  Justin Clark-Casey: I don't like adding more eyesores
[11:48]  Freaky Tech: have seen lots of projects that used that framework concept and that massive workarounds for getting parts to resolve circular dependencies
[11:48]  Justin Clark-Casey: because then when they need to be changed later there will be a lot of complaints about changes
[11:48]  Freaky Tech: if you have a good idea for moving modules, go ahead.
[11:48]  Marcus Llewellyn: I'm not remotely qualified to have an opinion on the elegance of OpenSim's internals, so I'm sorry if this is out of line. But if it comes down to a choice between waiting for a rearchitecture of opensim, or implementating a patch with immediate benefits, I'm kinda for the patch being accepted.
[11:49]  Hiro Protagonist: gotta bail ;) thanks to all you physics and vehicle devs ;)
[11:49]  Justin Clark-Casey: yes, that's why I think it's unavoidable in the shoirt term
[11:49]  Hiro Protagonist: o/
[11:49]  Aine.Caoimhe Marcus ++
[11:49]  Freaky Tech: I have never assumed, that it might stay where it is for all time.
[11:49]  AlexRime SiLiSiLi: We need buildoard of active regions with TP points in Plaza. Same as we have UserInWorld in Many newcommers even don't know about that page. It will connect active regions.
[11:49]  Hiro Protagonist: (thanks everyone else too :D)
[11:49]  Justin Clark-Casey: it's still an eyesore
[11:49]  BlueWall Slade: see ya Hiro
[11:49]  Marcus Llewellyn: B'bye Hiro :)
[11:49]  Freaky Tech: it was before Justin already
[11:50]  Dahlia Trimble: yes, changing stuff for asthetic reasons is not good because people downstream have dependent code
[11:50]  BlueWall Slade: Yeah, I think we shold take any good optimizations in with an eye for reorganizing and cleaning later.
[11:50]  Justin Clark-Casey: and then you can never refactor to actually allow a proejct to continue forward
[11:50]  Justin Clark-Casey: but we've been down this path before, I'm not interested in rehashing it now
[11:50]  AlexRime SiLiSiLi: OS demolish our office in Plaza with no reason at all. It is hurt bro
[11:51]  Freaky Tech: Justin you had missed several refactoring in the last years.
[11:51]  Freaky Tech: they are already making things hard to optimize
[11:51]  Justin Clark-Casey: ft: yes, and forks have done such massive changes
[11:51]  Justin Clark-Casey: with accompanying massive success
[11:52]  Dahlia Trimble: if you want to do major refactor, do it to opdate to a more modern architecture (hierarchical entity-component) rather than just for asthetic reasons
[11:53]  BlueWall Slade: ++
[11:53]  Freaky Tech: asthetics fixes are nice but do not bring a project a lot forward
[11:53]  Dahlia Trimble: and for every person who likes a certain asthetic design, theres others who prefer it another way
[11:54]  Dahlia Trimble: its very subjective
[11:54]  Freaky Tech: that is a real truth
[11:54]  BlueWall Slade wonders how al the asset name and description stuff ws never cleaned out when it started going to inventory.
[11:55]  Freaky Tech: refactoring is a process you need to do continuously
[11:55]  Dahlia Trimble: refactoring can also introduce regressions
[11:55]  Freaky Tech: sometimes a not so perfect thing according to asthetics has to be done at some point to get forward
[11:57]  Freaky Tech: defining an Interface and an implementation of it in the same .cs file is not really well structured at all
[11:57]  Freaky Tech: that is the case with HeightmapTerrainData anyways
[11:57]  Dahlia Trimble: defining an interface for a single implementation is probably not the best practice either
[11:58]  Freaky Tech: Dahlia I agree with that
[11:58]  BlueWall Slade: I guess much of our modules can be overwritten by third parties with that though.
[11:58]  Dahlia Trimble: maybe
[11:58]  Nebadon Izumi: hey Richardus, guess you didnt know about the time change eh?
[11:58]  Nebadon Izumi: DST strikes again
[11:58]  Dahlia Trimble: if the interface is well thought out, many arent tho
[11:59]  Richardus.Raymaker hi all
[11:59]  Freaky Tech: I did gothrough probably 60-70% of source code
[11:59]  Kayaker Magic: Daylight Wastings Time is such a waste of time....
[11:59]  Dahlia Trimble would prefer year long DST ;)
[12:00]  OtakuMegane Desu: ^
[12:00]  Justin Clark-Casey: alright, I'm off to take care of business. Goodbye everyone.
[12:00]  Dahlia Trimble: tc justin
[12:00]  Marcus Llewellyn: Take care Justin.
[12:00]  Freaky Tech: bye Justin
[12:00]  Aine.Caoimhe bye Justin
[12:00]  OtakuMegane Desu: bye
Personal tools
About This Wiki