Chat log from the meeting on 2024-01-23

From OpenSimulator

Jump to: navigation, search
[12:07 PM PT]  Ubit Umarov: yeack and welcome
[12:07 PM PT]  Ubit Umarov: starting with last week code changes..  only a few
[12:07 PM PT]  Ubit Umarov: i did review old warp3d map rendering code
[12:08 PM PT]  Ubit Umarov: for several regions, specially in windows i did reduce the peak memory usage a bit..
[12:08 PM PT]  Ubit Umarov: on this one, no idea why it did almost double :p
[12:09 PM PT]  Ubit Umarov: when code is a bit lighter.. does a bit less allocations.. ...  GC magic
[12:09 PM PT]  Ubit Umarov: final result should be the same uglyr images we get so far :)
[12:10 PM PT]  Ubit Umarov: another small cache was to add a option MaxOfflineIMs to robust offlineIM module
[12:10 PM PT]  Ubit Umarov: for some reason it had a hard limit of 25..
[12:10 PM PT]  Jamie.Jordan hi everybody
[12:11 PM PT]  Motoko.Karu hihi jamie
[12:11 PM PT]  Gavin.Hird 25 overall?
[12:11 PM PT]  Ubit Umarov: now should he possible to change on robust.ini
[12:11 PM PT]  Gavin.Hird Hi Jamie
[12:11 PM PT]  Ubit Umarov: per user
[12:11 PM PT]  Gavin.Hird oh, ok
[12:11 PM PT]  Ubit Umarov: user to user ims
[12:11 PM PT]  Gavin.Hird what is the new limit?
[12:11 PM PT]  Ubit Umarov: 25
[12:11 PM PT]  Ubit Umarov: :p
[12:11 PM PT]  Gavin.Hird :-)
[12:12 PM PT]  Ubit Umarov: but now you can change it on robust.ini
[12:12 PM PT]  Gavin.Hird I suppose that is a reasonable setting to avoid spamming
[12:12 PM PT]  Ubit Umarov: yeah..  should be kept low
[12:12 PM PT]  Vincent.Sylvester Word of caution I have reports that higher limits, say over 200 or so, may cause offline IM to fail to deliver at all. Still looking into what could be causing that. Probably have to change the delivery system a bit to make it run more reliably
[12:12 PM PT]  Ubit Umarov: well 200 is near suicidal ;)
[12:13 PM PT]  Vincent.Sylvester It is, but you know how it is, give em options and they'll run with em
[12:13 PM PT]  Ubit Umarov: well i did not added hard limits on that ( as i did on other options )
[12:13 PM PT]  Gavin.Hird brb
[12:14 PM PT]  Vincent.Sylvester 100 is probably fine, most reports I got I checked the table and saw over 300 pending IMs
[12:14 PM PT]  Ubit Umarov: remember that a user gets all that at arrival
[12:14 PM PT]  Ubit Umarov: bad karma
[12:14 PM PT]  Ubit Umarov: t have a big flood
[12:15 PM PT]  Vincent.Sylvester They weren't deleted either so that narrows down where the code might be failing, but... it doesn't throw an error of overflow or anything so it makes no sense why it would fail there
[12:16 PM PT]  Vincent.Sylvester Only thing I could think of is to deliver first 100 and then check if there are some left and run the routine again, not try the whole thing at once
[12:16 PM PT]  Ubit Umarov: they may point to a fail on get from the db
[12:16 PM PT]  Gavin.Hird back
[12:16 PM PT]  Ubit Umarov: since there is a delete all jsut after it
[12:17 PM PT]  Vincent.Sylvester That's what I been thinking too, data size exceeded on the fetch, but max packet on db is set to 1gb, shouldn't be that much data
[12:17 PM PT]  Ubit Umarov: you need to debug that :)
[12:17 PM PT]  Vincent.Sylvester fun stuff
[12:18 PM PT]  Ubit Umarov: ahh another change, Misterblue did revert a bullet lib to a previus version by default
[12:18 PM PT]  Ubit Umarov: he can explain :)
[12:19 PM PT]  Vincent.Sylvester That is a weird change. Someone running master dev code and yet an OS version that is coming up on being end of life...
[12:19 PM PT]  Ubit Umarov: hmm?
[12:19 PM PT]  Vincent.Sylvester Well someone complained on mantis about the lib not working because their ubuntu was older than what the lib was compiled with
[12:19 PM PT]  Misterblue Waves: I reverted the version in the repository since people were complaining about it not being compatible with older OS's
[12:20 PM PT]  Misterblue Waves: the new version is compiled on Ubuntu 22.04 which doesn't seem to be very downward compatible
[12:20 PM PT]  Vincent.Sylvester Yet they would only know if they ran master code, so they keep up with that, but don't bother to update their ubuntu at all
[12:20 PM PT]  Cuga.Rajal so issue was Linux only?
[12:20 PM PT]  Misterblue Waves: : I built it on Ubuntu 18.04 and 20.04. The 18.04 version is problematic since it's not supported any more but that seems to be the most compatible version
[12:20 PM PT]  Ubit Umarov: hmm you did also changed that lib when you added the others
[12:21 PM PT]  Misterblue Waves: The BulletSim binaries now have names that include their version and dates and stuff
[12:21 PM PT]  Cuga.Rajal If building on 18.04 makes it work on everything, that is proplly the best option
[12:21 PM PT]  Misterblue Waves: The old "BulletSim.dll/so" is changed for those people who don't update INI and .config files :)
[12:21 PM PT]  Cuga.Rajal Doesn;t matter that 18.04 is no longer supported
[12:22 PM PT]  Vincent.Sylvester 18 will be or already is end of life however, no one should still be using that unless they pay for extended security from canonical
[12:22 PM PT]  Ubit Umarov: "no longer supported" does not mean much in some cases
[12:23 PM PT]  Ubit Umarov: where ppl have nice machines running on a particular version and have no good reason to "update" into widl unknown bugs
[12:23 PM PT]  Cuga.Rajal I think Ubit was renaming the Mac binaries so not to include version and date
[12:23 PM PT]  Ubit Umarov: damm insane forced obsolecence...  whatever
[12:24 PM PT]  Vincent.Sylvester 18 is 6 years old now, 20 is 4 years old, by now I should think they have the most annoying bugs worked out
[12:24 PM PT]  Gavin.Hird They seemed unchanged, yes
[12:24 PM PT]  Ubit Umarov: i didn't change on this last ones
[12:24 PM PT]  Joe Magarac: 24 LTS Ubuntu should e out this summer. Then it's time to drop 18 LTS support, probably.
[12:24 PM PT]  Ubit Umarov: but we where suposed to have only one...
[12:24 PM PT]  Ubit Umarov: to ship..
[12:25 PM PT]  Cuga.Rajal might be better to leave name and date in those Bullet libs and update config
[12:25 PM PT]  Ubit Umarov: several is a extra complexity on config
[12:25 PM PT]  Misterblue Waves: it's not so much 18.04 being supported. BulletSim just needs libraries that re upward compatible
[12:25 PM PT]  Ubit Umarov: fun.. when bullet is pretty old now
[12:25 PM PT]  Cuga.Rajal do the Bullet libs built on 18.04 work on 22?
[12:26 PM PT]  Misterblue Waves: unless we are worried about legacy library security problems in OS (lol) it's not a problem
[12:26 PM PT]  Ubit Umarov: all recent updates where on the pything thing
[12:26 PM PT]  Ubit Umarov: python.. not on the main physics code
[12:27 PM PT]  Misterblue Waves: I haven't done enough cross version testing (like running the 18.04 on 22.04) but that is my plan before checking in the new version as the default one
[12:27 PM PT]  Vincent.Sylvester It's more a thing about enabling someone to run an unsupported version of ubuntu any longer when there are known security vulnerabilities that need patches, like the recent openssl thing in ssh. It just isn't such a great idea to not update a system.
[12:28 PM PT]  Vincent.Sylvester Plus it's just odd to keep up to date with master yet not bother to give the same attention to the operating system, maybe that's just me then
[12:28 PM PT]  Gavin.Hird agreed
[12:28 PM PT]  Ubit Umarov: well we are sutck on dotnet6
[12:28 PM PT]  Vincent.Sylvester Don't have to go bleeding edge on that, I stay a version behind on that too, but not 4
[12:28 PM PT]  Ubit Umarov: that does impose some conditions on the os alos
[12:29 PM PT]  Cuga.Rajal Is there an official list of OS versions supported for Opensim?
[12:29 PM PT]  Ubit Umarov: dotnet7 and 8 are a disaster..  GC is a total carp
[12:29 PM PT]  Ubit Umarov: crap
[12:29 PM PT]  Ubit Umarov: and ofc will force even more recent machines..
[12:30 PM PT]  Ubit Umarov: thin its dotnet6 list of supported machies
[12:30 PM PT]  Ubit Umarov: think..
[12:30 PM PT]  Misterblue Waves: I don't think there should be BulletSim binaries for each of the different Ubuntu release (and Debisn releases, ...)
[12:30 PM PT]  Ubit Umarov: well BigEndian cpus may not work well ( thing none is suported by dotnet )
[12:30 PM PT]  Cuga.Rajal agree
[12:30 PM PT]  Ubit Umarov: nahh we should only have one per platform
[12:31 PM PT]  Ubit Umarov: ie os   cpu
[12:31 PM PT]  Misterblue Waves: disitributing Linux binaries seems to be tricky
[12:31 PM PT]  Ubit Umarov: not per os version..
[12:31 PM PT]  Ubit Umarov: well a old problem
[12:31 PM PT]  Misterblue Waves: os X arch
[12:32 PM PT]  Ubit Umarov: ode is also compiled on older linux to avoid issues
[12:32 PM PT]  Ubit Umarov: its code does not use much fancy things that keep changing
[12:32 PM PT]  Misterblue Waves: if it works for ODE, it will work for BulletSim :)
[12:32 PM PT]  Ubit Umarov: well not if bullet uses fancy c   things
[12:32 PM PT]  Ubit Umarov: those change all the time
[12:33 PM PT]  Joe Magarac: Question: Is there anything special about "Friggin Awesome NE" and "Friggin Awesome E"? Firestorm sometimes gets stuck walking across that region boundary and Sharpview fails every time. All the right handshakes take place but the gaining region doesn't respond to AgentUpdate.
[12:34 PM PT]  Ubit Umarov: what are those Friggin ? :p
[12:34 PM PT]  Joe Magarac: Regions on OsGrid.
[12:34 PM PT]  Ubit Umarov: never been there
[12:34 PM PT]  Vincent.Sylvester Possible they are using ports that are problematic, who owns those regions?
[12:35 PM PT]  Joe Magarac: They have nice flat region edges.
[12:35 PM PT]  Joe Magarac: So I was testing there.
[12:35 PM PT]  Joe Magarac: Where's a good single region crossing edge that's land, not water?
[12:35 PM PT]  Misterblue Waves: I otta here... see you all later
[12:36 PM PT]  Ubit Umarov: you need to test things on well defined regions
[12:36 PM PT]  Joe Magarac: Ubittest1 and 2 are islands.
[12:36 PM PT]  Vincent.Sylvester How about I set you up a few regions to test with and give you ssh access to them so you can check their consoles for errors, cause without that you won't know why they failed
[12:36 PM PT]  Joe Magarac: I would appreciate taht.
[12:36 PM PT]  Ubit Umarov: as we told you osgrid has regions hen run on all kind of systems
[12:36 PM PT]  Ubit Umarov: and evne diferent versions
[12:36 PM PT]  Joe Magarac: Right.
[12:36 PM PT]  Joe Magarac: I think Friggin Awesome is on a nonstandard version.
[12:36 PM PT]  Ubit Umarov: all regions are islands.. what you mean?
[12:37 PM PT]  Joe Magarac: Land out to the edge of the region and a flat surface at the crossing.
[12:37 PM PT]  Joe Magarac: Not some big elevation jump and not underwater.
[12:38 PM PT]  Ubit Umarov: well there is land under water
[12:38 PM PT]  Ubit Umarov: but yeah it is flat
[12:38 PM PT]  Ubit Umarov: even so.. how did crossings worked there for you?
[12:38 PM PT]  Joe Magarac: All the handshakes work.
[12:39 PM PT]  Joe Magarac: But the gaining region doesn't seem to respond to AgentUpdate.
[12:39 PM PT]  Joe Magarac: I see the agent object created in the new region.
[12:39 PM PT]  Joe Magarac: I start sending AgentUpdate to the new region.
[12:39 PM PT]  Joe Magarac: But the new region doesn't send me ObjectUpdate.
[12:39 PM PT]  Joe Magarac: there may be bugs on my side.
[12:39 PM PT]  Ubit Umarov: regions only look to it when it makes sense
[12:39 PM PT]  Joe Magarac: Works fine for SL, though.
[12:40 PM PT]  Ubit Umarov: and they even do throttle it
[12:40 PM PT]  Joe Magarac: I can walk around SL freely. Even drive.
[12:40 PM PT]  Ubit Umarov: it does end objectupdate or we would not be here :)
[12:40 PM PT]  Joe Magarac:
[12:41 PM PT]  Joe Magarac: I'm probably doing something wrong. Most of my bugs at present are where I want a sim to send me something and it's no sending, but I don't know why.
[12:41 PM PT]  Joe Magarac: I appreciate Vincent's offer of help.
[12:42 PM PT]  Joe Magarac: Poking the black box is frustrating.
[12:42 PM PT]  Ubit Umarov: (Friggin Awesome NE is actually running recent 0.9.3 and on bullet )
[12:42 PM PT]  Joe Magarac: OK, so that is a reasonably good version of OS?
[12:43 PM PT]  Lyr Lobo smiles
[12:43 PM PT]  Joe Magarac: Are E and NE on the same version?
[12:43 PM PT]  Joe Magarac: I just had Firestorm get stuck at that crossing this morning.
[12:43 PM PT]  Joe Magarac: Just wlaking across.
[12:44 PM PT]  Ubit Umarov: yeah a bit of lag on crossings there
[12:44 PM PT]  Ubit Umarov: fun on the e > ne
[12:45 PM PT]  Ubit Umarov: ne -> seems better
[12:45 PM PT]  Ubit Umarov: ne -> e
[12:45 PM PT]  Joe Magarac: I see trouble with e->ne too.
[12:45 PM PT]  Ubit Umarov: well bullet crossings never did work well..  dunno
[12:46 PM PT]  Joe Magarac: Where should I test that should work? Ubittest1 -> 2 across the ocean bottom?
[12:46 PM PT]  Joe Magarac: (Sharpview is walk only right now. No flying.)
[12:46 PM PT]  Ubit Umarov: for example :)
[12:47 PM PT]  Joe Magarac: I will try that this afternoon. Thanks.
[12:47 PM PT]  Ubit Umarov: for now simple fat surfaces
[12:47 PM PT]  Joe Magarac: right.
[12:47 PM PT]  Ubit Umarov: flat..
[12:47 PM PT]  Gavin.Hird are you lloing for an intersection of 4 regions completely flat?
[12:47 PM PT]  Gavin.Hird looking*
[12:47 PM PT]  Joe Magarac: That's the next phase, yes.
[12:47 PM PT]  Ubit Umarov: i can edit the terrain on some spots if you need
[12:47 PM PT]  Joe Magarac: Thanks.
[12:48 PM PT]  Joe Magarac: I'm willing to test anywhere I can walk.
[12:48 PM PT]  Gavin.Hird what do you need, because I have some flat underwater
[12:48 PM PT]  Joe Magarac: right now, a single flat region crossing.
[12:48 PM PT]  Joe Magarac: Soon, four corners and the ability to rez vehicles.
[12:48 PM PT]  Gavin.Hird ok, hang on a sec
[12:50 PM PT]  Ubit Umarov: ( ofc a reason why we added large regions was bad crossings ;) )
[12:50 PM PT]  Joe Magarac: I know.
[12:50 PM PT]  Joe Magarac: but they should work.
[12:51 PM PT]  Joe Magarac: I have a lot of logging and can see region crossings in detail.
[12:51 PM PT]  Ubit Umarov: and they do..  if you pray well ;)
[12:51 PM PT]  Orbert.Tatham SL had the worst of both worlds last I tried sailing there
[12:51 PM PT]  Gavin.Hird If you go to the SE corner of the Firat region, it is reasonably flat and a road there. It is actually a 4 region crossing
[12:52 PM PT]  Gavin.Hird
[12:52 PM PT]  Gavin.Hird you can also rezz vehicles in the oposite corner on the pier
[12:52 PM PT]  Joe Magarac: Is that on OSgrid?
[12:52 PM PT]  Gavin.Hird It is my but you can just HG in there
[12:53 PM PT]  Joe Magarac: right now I only support OsGrid and Kitely.
[12:53 PM PT]  Gavin.Hird it runs pretty much the very latest update always
[12:53 PM PT]  Joe Magarac: It is my?
[12:53 PM PT]  Gavin.Hird so you can't hypergrid anywhere
[12:53 PM PT]  Joe Magarac: I don't have the grid-adding thing in Sharpview yet.
[12:54 PM PT]  Gavin.Hird ok, but you don't need that to HG teleport
[12:54 PM PT]  Joe Magarac: Right.
[12:54 PM PT]  Joe Magarac: I will try shortly. Thanks.
[12:54 PM PT]  Gavin.Hird sure
[12:55 PM PT]  Gavin.Hird if it works for you I can make a rezz zone at that corner
[12:56 PM PT]  Joe Magarac: Thanks.
[12:56 PM PT]  Kayaker Magic: I have a different subject: I'm working on a project where I have code writing LSL code and I ran into a limit: LSL scripts much larger than 262,000 characters cause strange problems and cannot compile/run. I assume I have run into a fixed limited size buffer in the viewer? 2 to the 18 characters maximum for any script?
[12:58 PM PT]  Gavin.Hird why the viewer, and not the script engine compiler?
[12:58 PM PT]  Kayaker Magic: Because the viewer often balks at large scripts and replaces them with a few gibberish characters.
[12:59 PM PT]  Gavin.Hird who have that large scripts anyway? :-)
[12:59 PM PT]  Ubit Umarov: script sources have a size limit
[12:59 PM PT]  Ubit Umarov: viewer side
[1:00 PM PT]  Kayaker Magic: dumb generated code tends to be large...
[1:00 PM PT]  Joe Magarac: Thanks all, got to go to next meeting...
[1:00 PM PT]  Cuga.Rajal take care Joe
[1:00 PM PT]  Vincent.Sylvester I don't want to even imagine the performance impact of that yikes
[1:00 PM PT]  Kayaker Magic: I was able to split the code into separate chunks and proceed.
[1:00 PM PT]  Ubit Umarov: think it used to be 64k not sure now
[1:00 PM PT]  Kayaker Magic: This is code that runs once, generates somethng then never runs again.
[1:01 PM PT]  Ubit Umarov: and that include the mess the precompiler or whatever adds
[1:02 PM PT]  Gavin.Hird I can check if there is a limit, but I'd be surprised if it is larger than 256k characters
[1:03 PM PT]  Ubit Umarov: Maximum script source code size - 65536 single byte characters (that's a viewer limit and can be changed in the config file 'panel_script_ed.xml').
[1:03 PM PT]  Ubit Umarov: on sl specs
[1:06 PM PT]  Gavin.Hird if that is the limit there certainly is a limit on storage as well
[1:07 PM PT]  Ubit Umarov: and fear fs preprocessor doubles the needed space..
[1:07 PM PT]  Ubit Umarov: well dunno never used it
[1:07 PM PT]  Object: Script running
[1:08 PM PT]  Object: Script running
[1:08 PM PT]  Kayaker Magic: I like the pre-processor, but yeah it doubles the source code size and I've been meaning to hack that out on 'finished products"
[1:09 PM PT]  Ubit Umarov: any other issue?
[1:09 PM PT]  Cuga.Rajal I have a dialog going with beq on the htauth issue
[1:09 PM PT]  Cuga.Rajal maybe something new to report soon
[1:10 PM PT]  Cuga.Rajal hgauth*
[1:10 PM PT]  Ubit Umarov: whatever that is :p
[1:10 PM PT]  Jagga Meredith: weird problem - I've got a space the size of this table I'm blocked from crossing.  Doesn't appear to he an object.  Ideas?
[1:10 PM PT]  Cuga.Rajal the 3-HTTP requerst thing on HG TP
[1:11 PM PT]  Ubit Umarov: like a "ghost prim" ?
[1:11 PM PT]  Jagga Meredith: yeah
[1:11 PM PT]  Ubit Umarov: well think we don't have those in years
[1:12 PM PT]  Ubit Umarov: and should vanish on restart..
[1:12 PM PT]  Ubit Umarov: if a delete fails
[1:12 PM PT]  Kayaker Magic: Gota run, RL calling...
[1:12 PM PT]  Jagga Meredith: OK, I'll restart
[1:12 PM PT]  Vincent.Sylvester In the developer menu under render metadata you can turn physics shapes on if you want to know what you are bumping into
[1:12 PM PT]  Lyr Lobo: bye Kayaker
[1:12 PM PT]  Jagga Meredith: k
[1:13 PM PT]  Cuga.Rajal Vincent will that work for physics objects too?
[1:13 PM PT]  Ubit Umarov: during delete, a object may stay doing collisions invisible for sometime while the delete actually happens
[1:13 PM PT]  Ubit Umarov: anything else for today? :)
[1:13 PM PT]  Jagga Meredith: ah ok.
[1:14 PM PT]  Gavin.Hird nope, I have to get back too
[1:14 PM PT]  Lyr Lobo: bye gavin
[1:14 PM PT]  Jagga Meredith: off to try it
[1:14 PM PT]  Ubit Umarov: ok thanks all. Hope to see you next week :)
[1:14 PM PT]  Gavin.Hird see you next week
[1:14 PM PT]  Cuga.Rajal me too
[1:14 PM PT]  Jamie.Jordan have a good week yall
[1:14 PM PT]  Cuga.Rajal take care all
Personal tools
About This Wiki