Chat log from the meeting on 2016-07-26

From OpenSimulator

Revision as of 13:18, 26 July 2016 by Sheera Khan (Talk | contribs)

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

[11:02] Sheera Khan: I have got a question...
[11:03] Sheera Khan: is there any way to see the memory usage of opensim instances a bit more detailed?
[11:04] Sheera Khan: like what amount is used by script, by objects, by avatars and so on...?
[11:04] Ubit Umarov: nopes
[11:04] Andrew.Hellershanks Sheera, it would be nice but I could not see any command to do that.
[11:04] Sheera Khan: at the console I mean
[11:04] Ubit Umarov: and its not that easy to add
[11:04] Robert Adams: I think what .NET/mono reports is what is displayed... would have to go into to the GC diagnostic calls to get more details
[11:05] Sheera Khan: we found the problem we had last week Andrew :-)
[11:05] Sheera Khan: it was a defective mesh object
[11:05] Sheera Khan: it sucked up several GBytes of RAM!
[11:06] Andrew.Hellershanks Sheera, keep a copy of that bad script. It can be looked at to see what was going on. A script shouldn't be able to suck up that much RAM.
[11:07] Sheera Khan: ty Robert, I'll dive a bit into that diagnostics...
[11:07] Ubit Umarov: ( she said mesh :) )
[11:07] Sheera Khan: it was a mesh object
[11:07] Sheera Khan: not a script
[11:07] Andrew.Hellershanks oh, yeah. Mesh object. That's a different issue.
[11:08] Andrew.Hellershanks That would make it more difficult to determine what was happening and if there is a way to stop it happening in the future.
[11:08] Sheera Khan: unfortunatly :-(
[11:09] Sheera Khan: but some kind of diagnostics might have helped a bit...
[11:10] Ubit Umarov: yeap diag is good
[11:10] Andrew.Hellershanks In Linux you can use valgrind to help with memory use analysis but I don't know how that would work with a program run via mono
[11:10] Ubit Umarov: but to a point it starts reducing performance
[11:11] Ubit Umarov: in same places it may already happen that stats/diag takes more time that the actuall job
[11:12] Ubit Umarov: and that is kinda of a nonsense
[11:12] Sheera Khan: sure, but in cases of looking for trouble a bit of diag would be helpful
[11:12] Sheera Khan: even if it slows the simulator down
[11:12] Sheera Khan: we even considered running it in a debugger ^^
[11:13] Ubit Umarov: thats what is needed in many cases, no matter the diag code added
[11:15] Sheera Khan: I'm really glad Mantis and Git show quite some progress in the last weeks :-)
[11:16] Andrew.Hellershanks yes. Development activity has definitely taken an upturn
[11:16] Sheera Khan: is there some sort of an ETA for a 0.9-final?
[11:16] Andrew.Hellershanks You would think the developers would want to be going outside in the nice summer weather instead of staying indoors and coding.. :)
[11:17] Andrew.Hellershanks There is talk of an 0.9 release this summer.
[11:17] Basil Sosides: wow
[11:17] Andrew.Hellershanks So, sometime in the next couple of months (approximately)
[11:17] Andrew.Hellershanks But we all know what those estimates of release dates are worth.
[11:18] Ubit Umarov: forgot to change to the alt name on a new viewer connection :)
[11:18] George Equus: Rather wait and make it as good as can than hurry just for the sake of a deadline date
[11:19] Gavin.Hird agreed George
[11:19] Sheera Khan: that's right though
[11:19] Andrew.Hellershanks A lot of items were on the 0.9 ToDo list on the wiki. Most have been dealt with.
[11:19] Andrew.Hellershanks Just 9 still on the list and one of them I've tested and believe is fixed.
[11:20] Ubit Umarov: ( and new bugs added :p )
[11:20] Sheera Khan: and pgSQL is catching up too :-)
[11:20] Gavin.Hird that too
[11:20] Andrew.Hellershanks yes. There was some other PgSQL issue I noticed in the IRC channel the other day
[11:21] Gavin.Hird what was that Andrew?
[11:21] Andrew.Hellershanks You expect me to remember something from yesterday?  ;)
[11:21] Andrew.Hellershanks Let me look...
[11:21] Gavin.Hird :-))
[11:22] George Equus: what was the question again...
[11:22] Sheera Khan: hihihi - looking it up in extended Memory - aka chat-log
[11:22] Andrew.Hellershanks ah, yes. It was about offline messages not being sent/receive from/to PgSQL regions.
[11:22] Andrew.Hellershanks #7975
[11:23] Gavin.Hird yes
[11:23] Gavin.Hird that needs an update to PostgreSQl 9.5 + rewriting the generic table handler
[11:24] Gavin.Hird because the code relies on the MySQL extention replace into SQL statement
[11:24] Gavin.Hird and the equivalent only exist in 9.5
[11:24] Gavin.Hird or higher
[11:24] Andrew.Hellershanks Replace does insert or update as needed?
[11:25] Gavin.Hird yes
[11:25] Gavin.Hird a socalled upsert
[11:25] Andrew.Hellershanks Bad idea using something like that which might be unique to a particular DB
[11:25] Andrew.Hellershanks upsert?
[11:25] Andrew.Hellershanks upset?
[11:25] Sheera Khan: update + insert
[11:25] Gavin.Hird it is called upsert - combination of update and insert
[11:25] Andrew.Hellershanks oh. :)
[11:26] Andrew.Hellershanks I can see that as a useful feature if porability isn't an issue.
[11:26] Andrew.Hellershanks portability
[11:26] Gavin.Hird so there is a few areas where things don quite work because of that
[11:26] Andrew.Hellershanks nods
[11:27] Gavin.Hird offline IM is one of them
[11:27] Gavin.Hird online IM should work as expected
[11:28] Andrew.Hellershanks The report mentioned use of v1 so I thought it would be a case of an external PHP based script being involved to do the actual DB updates.
[11:29] Gavin.Hird ok, the php is most likely not written with PGSQl syntax
[11:29] Gavin.Hird which is different
[11:29] Andrew.Hellershanks The PHP script I use for updating the offline IM DB tables doesn't use Replace.
[11:30] Gavin.Hird no but the MySQL php syntax will not work
[11:30] Gavin.Hird the php must have all database related statements rewritten
[11:30] Andrew.Hellershanks It sounds more like a PHP issue rather than a core OS one.
[11:31] Gavin.Hird yes
[11:32] Andrew.Hellershanks Now we need one more person to sit between Ken and Ubit to have every other seat occupied. :)
[11:32] Ken Savage: It needs to be a girl considering the animation on that seat lol
[11:32] Andrew.Hellershanks well... I was tempted to ask where is Barbie  ;)
[11:34] Andrew.Hellershanks Perhaps its time to put Robert on the spot and ask how things are going with the physics engine(s)
[11:34] Robert Adams looks in inventory for a spot to stand on
[11:35] Andrew Hellershanks: You could probably find a spot lost in the shadow of the moon on the floor
[11:35] Robert Adams: haven't done much to BulletSim... there are still a bunch of bugs to fix and I'd like to move to Bullet3 (the new version with HW acceleration)
[11:36] Andrew Hellershanks: Robert, any show stopper type issues in BS?
[11:36] Robert Adams: If I had infinite time, I've been thinking of redoing how physics fits into the simulator but that is mostly dreaming
[11:37] Robert Adams: I've heard of some problems with vehicles and borders since the merge... the sensing of region edges must be different
[11:37] Andrew Hellershanks: Vehicles and borders has been an issue for a while.
[11:37] Robert Adams: other than that, some people think some of the outstanding bugs are showstoppers for them but I haven't dig into them
[11:38] Ubit Umarov: what issues ??
[11:38] Andrew Hellershanks: Robert, do you have mantis #'s for any of those possible show stopper issues?
[11:38] Robert Adams: I looked at the additional CPU usage by 090... that doesn't seem to be the physics engines... I'm suspecting some of the new work queue stuff but haven't confirmed
[11:39] Andrew Hellershanks: ok
[11:39] Ken Savage: Tell us more about Bullet 3 and how much time it might take to get it working here :)
[11:39] Andrew Hellershanks: That may limit the search for that issue a little bit.
[11:39] Ubit Umarov: but what border issues?
[11:39] George Equus: rubber banding?
[11:40] Ubit Umarov: crossings where not working before the merge now they do
[11:40] Robert Adams: 7779, 6782, 7303 7132 , 7768, 7862 and others
[11:40] Andrew Hellershanks: ok. That's quite a list.
[11:40] Ubit Umarov: rubber banding is not possible to reduce completly
[11:41] Basil Sosides: same as in SL
[11:41] Ubit Umarov: it also happens at sl and crossings are pretty good there now
[11:41] Andrew Hellershanks: Robert, for the purposes of tracking issues that could/should/must be addressed for 0.9, if you give me a list of them I can add them to the wiki page.
[11:41] Ubit Umarov: 7779 is not borders
[11:42] Ubit Umarov: and that that push also fails at sl if i remember
[11:42] Robert Adams: you're talking about the wiki page at ?
[11:42] Andrew Hellershanks: yes
[11:43] Robert Adams: I'll add ones I think are regressions.... I'd like to keep that the things to fix to make 090 as usable as 082
[11:43] Andrew Hellershanks: ok
[11:43] Ken Savage: Might your time be better spent on Bullet 3 or will the fixes crossover?
[11:43] Ubit Umarov: err what regressions ??
[11:44] Robert Adams sticks tongue out at Ubit
[11:44] Andrew Hellershanks: My suspicion is that B3 would need more time to get useable than the time to fix a few issues in BS.
[11:44] Ubit Umarov clips Robert tongue
[11:44] Robert Adams: any additions for integrating Bullet3 would be downward compatible
[11:46] Robert Adams: but it is hard to stay 100% SL compatible... a lot of that presumes a particular implementation
[11:46] Andrew Hellershanks: When we have a new OS specific viewer that won't be such an issue any more.
[11:46] Robert Adams: I'd like to intoduce an additional vehicle model in addition to the SL legacy vehicle... but that would be OpenSimulator specific
[11:47] Andrew Hellershanks: What features would be in the new model?
[11:47] Robert Adams looks in inventory for an OS specific viewer.... humm... some problem with inventory downloading completion ;-)
[11:47] Gavin.Hird LL seems to be waffling a bit at the moment on their own compatibility
[11:48] Andrew Hellershanks: Their level of support for Linux is dropping off. That makes things harder for TPV devs.
[11:48] Gavin.Hird but at least they seem to waffle with absolute conviction.
[11:48] Andrew Hellershanks: hehe
[11:49] Gavin.Hird they have basically dropped Linux support
[11:49] Andrew Hellershanks: I know. I'll have to stop using their viewer soon the way things are going.
[11:49] Ubit Umarov: that inventory issues sounds like a fs issue
[11:49] Ubit Umarov: works fine with singu
[11:50] Ken Savage: But sing has the issue of textures not rezing sometimes too, so both have issues
[11:50] Andrew Hellershanks: If LL has dropped support for Linux I wonder how much effort the TPV devs will be willing to put in to support that operating system.
[11:50] Ubit Umarov: i did notice FS even repeats same request on diferent fetchdesc call
[11:50] Gavin.Hird I can't say we know about any inventory downloaing isses for Kokua
[11:50] Ubit Umarov: that is a extra 1MB on my test case :)
[11:50] Ubit Umarov: and does that several times
[11:51] Gavin.Hird NickyP is pretty dedicated to keep a Linux version of Kokua running
[11:51] Ubit Umarov: today i was at sl looking for a fs dev.. guess ill need to file a jira
[11:51] Andrew Hellershanks: Ken, There have been a number of issues with recent versions of Sing. I haven't been using it as much any more.
[11:51] Ubit Umarov: fs also has an issue with a call card
[11:52] Ken Savage: Yea, I had to stop using sing because of the gray texture issues
[11:52] Ubit Umarov: singu 1.8.7 seems ok ( don't have show property lines on minimap )
[11:53] Ubit Umarov: singu 1.8.7 seems ok ( don't have show property lines active on minimap/radar )
[11:53] Gavin.Hird Kokua opensim builds are at
[11:53] Andrew Hellershanks: I've stopped using Sing because it is no longer very stable. It often just crashes on me for no apparent reason.
[11:53] Andrew Hellershanks: 1.8.7 is better than the Alphas I used to be using but it still isn't as stable as it used to be.
[11:54] Andrew Hellershanks: Gavin, where are crash report files saved by Kokua under Linux? I want to file a report about the viewer crashes I get.
[11:54] Ubit Umarov: see those lines on the minimap..
[11:55] Andrew Hellershanks: Gavin, it often doesn't like it when I switch desktop workspaces.
[11:55] Gavin.Hird I think we have disabled them
[11:55] Ubit Umarov: turn then off
[11:55] Simulator Version v0.5 ruft: OpenSim Dev 3f0d84c: 2016-07-25 19:49:40 +0100 (Unix/Mono)
[11:55] Ubit Umarov: that is fixed on the next fs and singu versions
[11:56] Ubit Umarov: well i use win versions.. linux may have other issues :)
[11:56] Gavin.Hird look for a file settings_crash_behavior.xml
[11:56] Gavin.Hird set the crash log varaible to 2 to make it stop generating / sending crash reports
[11:56] Andrew Hellershanks: Gavin, ok. ty
[11:57] Gavin.Hird it shall exist in the user_settings directory
[11:57] Andrew Hellershanks: Its default behaviour is to send reports? I thought only sing did that.
[11:58] Andrew Hellershanks: I've not noticed Kokua reporting that it was sending any report after a crash.
[11:58] Gavin.Hird it is an on and off issue, but new installations should have that file with a setting of 2
[12:00] Andrew Hellershanks: Gavin, found the file. I've changed the setting to 0
[12:00] Gavin.Hird ok
[12:01] Ken Savage flüstert: Is everyone having very slow AV rez issues here on osg for the last several weeks, or is it just me?
[12:01] Andrew Hellershanks: I mostly run a viewer in its own workspace. Having a viewer crash when I temporarily switch workspaces to do something else for a moment is very annoying.
[12:02] Gavin.Hird I am afraid we don't have a good solution for that right now
[12:02] Sheera.Khan Kokua_64_4_0_2_38137_x86_64.tar.txz ? What is txz for a compression?
[12:02] Gavin.Hird sounds like a compressed tar
[12:02] Andrew Hellershanks: txz? usually it would be xz which is xzip or possibly 7zip
[12:02] Andrew Hellershanks: gz would be gzip
[12:03] Sheera.Khan ahh, it is an actual xz-file
[12:03] Sheera.Khan just a typo I suppose
[12:03] George Equus: Not me Ken, seem OK, pretty instant here eg, bit slower on my regions (due to slow hardware)
[12:04] Sheera.Khan @Ken, I don't have that issue here...
[12:05] Andrew Hellershanks: Darn SourceForge. I have a wide screen monitor but can't see the proper file names on that Kokua web page because they limit the width of the column showing the available versions so they can stick in a short advert on the right side of the column.
[12:06] Gavin.Hird no adblocker?
[12:06] Andrew Hellershanks: I do but it didn't see that as an add that needs to be blocked.
[12:06] vegaslon plutonian: throw it in a rss veiwer
[12:07] Gavin.Hird adblocker + no flash installed + no java installed makes wonders
[12:07] Andrew Hellershanks: oh, sorry. Not an advert. Just "recommended projects"
[12:07] Sheera.Khan @Andrew: just above the table you'll see:
[12:07] Sheera.Khan Looking for the latest version? Download Kokua_64_4_0_2_38137_x86_64.tar.txz (77.0 MB)
[12:07] Andrew Hellershanks: that's the version I already have.
[12:07] Sheera.Khan that is the link you're looking for
[12:07] Andrew Hellershanks: I see what you mean about the typo
[12:08] Andrew Hellershanks: I have the latest version.
[12:08] Gavin.Hird
[12:08] Ken Savage: AV rez on my test grid is immediate, but on osg it now takes a minute or longer to rez
[12:09] Gavin.Hird they should be called KokuaOS and not just Kokua
[12:09] Sheera.Khan hmmmm: bin/do-not-directly-run-kokua-bin: error while loading shared libraries: cannot open shared object file: No such file or directory
[12:09] Sheera.Khan even though i | libffi4 | Foreign Function Interface library | Paket
[12:10] Sheera.Khan is installed
[12:10] Andrew Hellershanks: Other files listed on that page are .txz. Perhaps an error in the package creation script.
[12:10] Andrew Hellershanks: Sheera, I have libffi6 installed on my machine.
[12:11] Gavin.Hird
[12:12] Sheera.Khan oh, I got it extracted, tar did recognize it
[12:12] Sheera.Khan it was just me wondering what txz is...
[12:12] Gavin.Hird tar xf your_file.txz --use-compress-program xz
[12:13] Andrew Hellershanks: ok, txz is just xz in the same way that tgz is just gz.
[12:13] Andrew Hellershanks: Gavin, if you just use tar xf it is usually able to determine which program it needs to use to extract the files.
[12:14] Gavin.Hird right
[12:14] Gavin.Hird the command was just in case it didn't
[12:14] Andrew Hellershanks: As mentioned in that page you can add J to the options to force use of xz
[12:15] Andrew Hellershanks: It's after the hour and we already lost a few other visitors to the meeting. Any final thoughts for today before we head our separate ways until around this time next week?
[12:16] Gavin.Hird Nothing from me
[12:16] Sheera.Khan seems everyone is happy :-)
[12:16] Gavin.Hird thanks everyone!
[12:16] Andrew Hellershanks: well... Not so sure about that but...
[12:16] Andrew Hellershanks: :)
[12:17] Andrew Hellershanks: I'm going to get back to work on a program I'm working on for a microcontroller. See you all again next week.

Personal tools
About This Wiki