Chat log from the meeting on 2016-05-10

From OpenSimulator

Jump to: navigation, search

[11:02] Andrew.Hellershanks Here we are again. :)
[11:06] Sheera Khan: Are there any updates on the status of the merge of the Avination stuff?
[11:06] Sheera Khan: Is the merge done ?
[11:06] Sheera Khan: Are we "only" waiting for the dust to settle?
[11:06] Kayaker Magic: We are waiting for new developers to join the team who can fix the bugs reported.
[11:07] Gavin.Hird it is summer…
[11:07] Sheera Khan: waiting for new developers could take a bit ...
[11:07] Billy.Bradshaw Mantis 7900 with patch to improve stability when running opensim on linux later than 3.10. Any comments from the core developers?
[11:07] Andrew.Hellershanks The main part of the merge was done. It's mostly waiting for the dust to settle.
[11:08] Sheera Khan: There still are some showstoppers I've heard...
[11:08] Gavin.Hird is there anyone who can even pick up after ubode - anyone who knows the avination code?
[11:08] Kayaker Magic: Sorry, I was being cynical. I have a plan to try and become a developer myself and try to fix some of my (un)favorite bugs.
[11:09] Jeff.Kelley @ MANTIS 7903 has a patch to fix sit position
[11:09] Sheera Khan: that would be great Kayaker ;-)
[11:09] Andrew.Hellershanks Sheera, indeed. It can take a while to become familiar with the OS code base.
[11:09] Sheera Khan: I'd really love to see you as a core developer
[11:10] Sheera.Khan unfortunatly I only know about some C.. C++ and C# are well beyond me :-(
[11:10] Kayaker Magic: One idea I have is to look at some of the smaller bugs, like the incorrect return from llGetCameraRot, and compare it to the InWorldz code that does it correctly.
[11:10] Andrew Hellershanks: Sheera, I was told there aren't any known showstoppers in git master. What issues have you heard about or run across that are show stoppers?
[11:11] Kayaker Magic: Something small like that cannot have diverged far.
[11:11] Jeff.Kelley @ inability to create landmarks is a show-stopper
[11:11] Kayaker Magic: And that would be a way for me to get up to speed....
[11:11] Sheera.Khan Personally I don't run 0.9 as of yet as I wasn't aware that the merge is mostly done
[11:12] Gavin.Hird I'd say that any lost functionality from a previous version is a showstopper
[11:12] Sheera.Khan I had the impression there still was quite a bit in limbo ...
[11:12] Kayaker Magic: The inability for avatars to cross SIM borders in some situations is a show stopper.
[11:12] Kayaker Magic: in
[11:12] Sheera.Khan but Kay mentioned some showstoppers
[11:13] Andrew Hellershanks: Billy, I haven't seen the report until now. I'm not sure if any of the other core devs have so no comment on it yet.
[11:13] George Equus: I now have almost perfect trans from region to region on my sim. am on 0.9
[11:13] Kayaker Magic: I've been testing a lot, there is much to like about it.
[11:14] Billy.Bradshaw @Andrew It is especially important to smxy who included the patch to enable him to run on centos 7
[11:14] Andrew Hellershanks: Sheera, C# is close enough to see for the most part that I didn't find it that hard to do some work on the OS code. If you know C and C++ I don't think you would find it that bad.
[11:14] Gavin.Hird I find it mostly stable and as I fixed issues in the viewer it has improved
[11:14] Kayaker Magic: If you have two var regions next to each other, and they are on the same physical server, an avatar can only cross the border once, then gets stuck.
[11:15] Gavin.Hird but region crossings can be challenging and the high CPU use is an issue
[11:15] Kayaker Magic: (and they are the same size and are in separate instances)
[11:16] Andrew Hellershanks: Kayaker, what is the mantis # re: the problem with sim crossings?
[11:16] Kayaker Magic: Hold on, I'll look it up.
[11:17] Andrew Hellershanks: ok
[11:17] Andrew Hellershanks: Kayaker, The two regions are properly placed up against each other?
[11:17] Kayaker Magic: It is number 7862
[11:18] Andrew Hellershanks: Kayaker, ok, ty
[11:18] Kayaker Magic: Yes, against each other.
[11:18] Andrew Hellershanks: Same size and not vertically or horizontally offset from each other?
[11:18] Kayaker Magic: If I put both vars in the same instance, crossing works well.
[11:18] Kayaker Magic: And by the way, putting to vars in the same instance does work.
[11:19] Kayaker Magic: Although you wouldn't want to put big vars or lots of content together like that normally.
[11:20] Andrew Hellershanks: Gavin, high CPU usage during the sim crossing?
[11:20] Gavin.Hird all the time
[11:20] Andrew Hellershanks: What do you consider high?
[11:20] Gavin.Hird it is close to 50% higher than 0.8.2
[11:21] Kayaker Magic: For testing, I set up two 1x1s, two 2x2s two 3x3s and two 4x4s next to each other, could not have gotten it wrong in EVERY case....
[11:21] Gavin.Hird both with bullet and ubODE
[11:21] Simulator Version v0.5 ruft: OpenSim Dev 1e44aba: 2016-05-02 21:40:14 +0200 (Unix/Mono)
[11:21] Gavin.Hird for an idle grid that does not mean so much, but once it gets busyer it really adds up
[11:23] Andrew Hellershanks: Gavin, have you filed a report about the CPU load?
[11:23] Gavin.Hird no
[11:23] Andrew Hellershanks: ok.
[11:23] Gavin.Hird AFAIK the developer has left the house
[11:24] Billy.Bradshaw Would it not be best to wait for the current batch of reports to at least have some process before adding more?
[11:24] Andrew Hellershanks: I will ask about it and see if any of the other devs have noticed a problem with increased CPU usage.
[11:24] Sheera.Khan there still are some devs in the house, Andrew i.e. ;-)
[11:24] Kayaker Magic: Another show stopper for me is the llCastRay function in ubODE taking over a second to work in var regions.
[11:25] Sheera.Khan outch
[11:26] Kayaker Magic: I have several llCastRay mantises, 7851, 7850, plus some other ubODE mantises.
[11:26] Andrew Hellershanks: Kayaker, that will be an issue for Ubit to look at.
[11:27] Gavin.Hird the problem is that only ubit knows the avination code well enough to take action on it
[11:28] Gavin.Hird question: Does anyone know why someone like Melanie never is at these meetings?
[11:28] Andrew Hellershanks: Melanie may know that code too. ubODE was a project of Ubits
[11:29] Andrew Hellershanks: I don't know why other core devs are seldom at these meetings. I haven't even seen radams here in a while.
[11:31] Gavin.Hird I've been reading the minutes for probably a year and a half and there are rarely anyone of the developers present
[11:31] Kayaker Magic: Justin used to come to every meeting. I miss Justin :(
[11:32] Gavin.Hird he did
[11:32] Billy.Bradshaw And Dahlia, and Bluewall
[11:32] Andrew Hellershanks: yes, he was the one core member that had been here at most of the meetings.
[11:33] Gavin.Hird it probably feels too much like work, hehe
[11:33] Billy.Bradshaw I always valued Dahlia's input, maybe if we offer free chocolate cookies
[11:33] James.Atlloud There was a lot of recognition for Justin at the time he had to leave. Would more recognition be valuable WHILE a dev is participating?
[11:34] Andrew Hellershanks: James, couldn't hurt.
[11:34] James.Atlloud OK - thanks. Something to think about.
[11:34] Andrew Hellershanks: People are often quick to criticize but slow to praise.
[11:34] James.Atlloud So true.
[11:35] Billy.Bradshaw Maybe one way forward would be for outsiders to release binaries with fixes for the community to try
[11:35] Andrew Hellershanks: Billy, the patches applied to the binaries should be available at the same time for review.
[11:35] Gavin.Hird there has been done / contributed a LOT of good work on 0.9 dev. It is just the final sprint that is missing.
[11:36] Kayaker Magic: there is the exchange rate: One OH SHIT cancels 10 ATTA BOYs
[11:36] Andrew Hellershanks: Gavin, its typical of the 10/90 issue of coding.
[11:36] James.Atlloud lol @ATTA BOY
[11:36] Gavin.Hird yep
[11:37] Billy.Bradshaw Well I have tried a patch and there seems to be delays in acknowledging it let alone test it
[11:37] Gavin.Hird We'll see if the Plox scripting engine comes to frution now that the port has been funded
[11:37] Andrew Hellershanks: Gavin, What is that?
[11:37] Gavin.Hird the Inworldz scripting engine
[11:38] Gavin.Hird which is said to be very SL compatible
[11:38] Kayaker Magic: and very fast
[11:38] Kayaker Magic: and has a lot of bugs fixed
[11:38] Gavin.Hird they raised $2000 for the port
[11:38] Kayaker Magic: they?
[11:38] Andrew Hellershanks: Oh. The problem with that is that InWorldz has their own code base. It may not be easy to integrate in to main stream OS.
[11:38] James.Atlloud phlox
[11:38] Gavin.Hird people
[11:39] Gavin.Hird they said they can do it
[11:39] Gavin.Hird so hopefully
[11:39] Andrew Hellershanks: Its worth looking at. There has been talk of redoing the scripting engine.
[11:39] Gavin.Hird
[11:40] Kayaker Magic: A new script engine is a huge task, starting with Phlox is a huge leg up.
[11:40] Kayaker Magic: Even if it is not "compatible"
[11:40] Andrew Hellershanks nods
[11:40] Gavin.Hird does it support ossl?
[11:41] Kayaker Magic: It has its own set of similar functions...
[11:41] Gavin.Hird ok
[11:41] Andrew Hellershanks: I was already reading a page about it just now. It claims 100% compatability with SL. To do that isn't just a case of the scripting engine but support in the main part of the grid code to back up the scripting functions.
[11:41] Jeff.Kelley @ better drop LSL if we are heading to a new script engine
[11:42] Billy.Bradshaw That could get messy, if scripts from one grid blow up on another
[11:42] Gavin.Hird In other news LL has said they will start to turn off legacy communication (UDP) so older viewers may stop being able to work in SecondLife
[11:42] Gavin.Hird Singularity is probably the most vulnerable
[11:42] Billy.Bradshaw I use C# features exclusively on our internal grid
[11:43] Andrew Hellershanks: Gavin, I have been sticking to the LL viewer when I'm in SL. I haven't used a third party viewer in that grid in ages.
[11:43] Kayaker Magic: I run most of my scripts in InWorldz (Phlox) and OpenSim (xengine) it is possible to work in both.
[11:44] Andrew Hellershanks: Billy, switching to a different scripting engine is a big deal. There will be a massive number of complaints if a new engine breaks scripts.
[11:44] Andrew Hellershanks: Kayaker, one issue I have with InWorldz is they don't preserve prim link order.
[11:45] Gavin.Hird no, but as I said before almost all viewers depend on the LL code, and they have been "sneaking in the backdoor" code changes for a long time that is quite incompatible with OpenSim. Suddenly all communication with the simulators and grid is http based.
[11:45] Andrew Hellershanks: They use an old way of combining linksets.
[11:45] Gavin.Hird so to match that quite substantial work will be needed on the opensim server
[11:45] Billy.Bradshaw There is a new inworld script editor with proper errors and line numbers, but that seems to gone cold as well
[11:47] Andrew Hellershanks: Another victim of the 10/90 coding issue?
[11:47] Billy.Bradshaw Well one good thing, opensim is really about content imo, and if that can transfer who cares what runs it
[11:48] James.Atlloud content = assets?
[11:48] Billy.Bradshaw yes
[11:48] James.Atlloud ty
[11:48] Billy.Bradshaw I mean we have HG for grids to communicate
[11:48] Gavin.Hird the new avatar skeleton cannot transfer and the animations that is developed for it
[11:49] Gavin.Hird at least not now
[11:49] Andrew Hellershanks: Gavin, Cannot transfer from where to where?
[11:49] Gavin.Hird the new skeleton and animations needs server side support
[11:50] Billy.Bradshaw And open source philosophy is if it don't do what you want you can change it
[11:50] Gavin.Hird so anything developed for it will not import to opensim
[11:51] Gavin.Hird you can actually upload it, but the result is disastrous as far as avatar geometry :-))
[11:51] James.Atlloud flüstert: One can only imagine it is like a transporter accident
[11:52] Billy.Bradshaw I already adjust width and breadth of an avatar from the visual params, to make things like animals collision more effective
[11:53] Billy.Bradshaw I believe that is what LL have done for some time
[11:53] Gavin.Hird they have added bones for the fingers, for the face, for wings, for tails
[11:53] Billy.Bradshaw yes
[11:54] Kayaker Magic: whoa!
[11:54] Gavin.Hird so facial expressions can be animated and not being morph targets on the avatar mesh
[11:54] Andrew Hellershanks: That's going to require updates to both backend code and the viewer.
[11:55] Gavin.Hird viewer 5 will be launched soon with full support for it
[11:55] Gavin.Hird now they have frozen the rigging
[11:56] Gavin.Hird but probably only with the ability to connect to the backend over http
[11:56] Billy.Bradshaw Gavin, are you saying they will stop using UDP for avatar positioning updates?
[11:56] Gavin.Hird yep
[11:57] Billy.Bradshaw I think that's nuts
[11:57] Gavin.Hird nuts or not that is what they have been cooking since mid 2014
[11:58] James.Atlloud Are games in general shifting from UDP?
[11:58] Billy.Bradshaw James, not as far as I know, quite the oposite
[11:59] James.Atlloud ok - ty.
[11:59] Billy.Bradshaw I did have a list of recent papers on the subject, cannot find at the moment.
[11:59] Andrew Hellershanks: The hours is about up. Any other final thoughts or comments for this week?
[11:59] James.Atlloud Maybe a shift to http is better for mobile devices?
[11:59] Andrew Hellershanks: There would be a little less overhead with UDP
[12:01] Billy.Bradshaw the rest nature of http seems to make any kind of synchronization very difficult. but who knows
[12:01] Gavin.Hird I think it also has to do with many ISPs and organizations blocking UDP
[12:01] James.Atlloud oh yes, there's that. uff
[12:01] Kayaker Magic: synchronization works in OpenSim now???!
[12:02] Andrew Hellershanks: As long as there is a timestamp in the packets it would be possible to keep things in sync.
[12:02] Billy.Bradshaw Kayaker over UDP yes
[12:03] Andrew Hellershanks: Gavin, some ISPs are trying to control traffic generated by peer to peer programs often used for file sharing.
[12:04] PcTek CyberStar: well udp comes from a place
[12:04] PcTek CyberStar: and it's not really traceable backwards
[12:04] PcTek CyberStar: udp is i think faster b/c it doesn't have the overhead of TCP
[12:05] Gavin.Hird udp just blast packets and hopes they reach someone
[12:05] PcTek CyberStar: right but tcp checks
[12:05] Gavin.Hird TCP has sequence numbers and can be reassembled and retransmitted in the correct order
[12:05] PcTek CyberStar: and leaves a 2 way street
[12:06] Andrew Hellershanks: Gavin, Correct.
[12:06] Gavin.Hird yes :-))
[12:07] PcTek CyberStar: one thing i learned about sL hanging on.
[12:07] Andrew Hellershanks: I'm going to have a go at shaking things up with the other core devs to get some answers about a few of the outstanding issues and the state of OS development.
[12:07] PcTek CyberStar: was you can run 3 copies of mtr in linux
[12:07] PcTek CyberStar: and point them to different ip's in sL
[12:07] PcTek CyberStar: of 3 good running sims
[12:08] PcTek CyberStar: and the router ai over the net favourites your path in cache
[12:08] PcTek CyberStar: so your ride to sL would be super smooth
[12:08] PcTek CyberStar: and all packet loss would slowly decrease to nothing
[12:08] PcTek CyberStar: even if it was generated by your isp
[12:08] PcTek CyberStar: they couldn't prise you off sL if you use that technique
[12:08] PcTek CyberStar: of course on windows you have to use 3 copies of winmtr
[12:09] PcTek CyberStar: sort of my anti-lag hacker tip
[12:10] Andrew Hellershanks: PCTek, How would you get the different IPs in SL?
[12:10] Billy.Bradshaw Gavin any specific references to the SL viewer plans you know of?
[12:10] Gavin.Hird they slip it out in small pieces here and there
[12:10] Gavin.Hird their Bitbucket is buzzing at the moment with code changes and projects
[12:11] Billy.Bradshaw Gavin is version 5 on their bucket site?
[12:11] Billy.Bradshaw lol bitbucket
[12:11] Gavin.Hird yes
[12:11] Gavin.Hird
[12:11] Gavin.Hird viewer bento
[12:11] Billy.Bradshaw kk, I find the notes in their source useful
[12:11] Gavin.Hird or bento-box as the project is called
[12:11] PcTek CyberStar: they should code something into their simulators to stop griefers from ddosing the sims when they get mad.
[12:11] Billy.Bradshaw thks Gavin
[12:11] Sheera.Khan I'd soooo appreciate Bento ported to OpenSim ^^
[12:12] Gavin.Hird that would be fantastic :-)
[12:12] Sheera.Khan finally having real bones to move a tail ;-)
[12:12] James.Atlloud Yes, this would be very popular.
[12:12] Billy.Bradshaw Are there still rules as to people working on both viewer and simulator code.. do we need dirty developers :)
[12:12] Gavin.Hird and there is a very long discussion running in their forums
[12:12] Sheera.Khan and thus being able to animate them and express emotions ...
[12:13] Gavin.Hird
[12:13] vegaslon plutonian: bento will come in time, they have not even officially finalized the new skeleton.
[12:13] Andrew Hellershanks: Billy, the restriction about not being able to work on OS code if you have seen viewer code was dropped ages ago.
[12:13] Gavin.Hird they have now
[12:13] vegaslon plutonian: they have put that in their bento release notes a couple of times and still changed the skeleton
[12:15] Billy.Bradshaw I am only here for the code side of things, don't do any building. so sorry if I don't always appreciate their concerns
[12:15] Andrew Hellershanks: Time to call this meeting concluded for this week. Good seeing everyone. I'll try and get some answers to the issues raised for next week.
[12:15] Andrew Hellershanks: I keep getting sidetracked with a number of other projects.

Personal tools
About This Wiki