Chat log from the meeting on 2014-11-25

From OpenSimulator

Revision as of 17:15, 26 November 2014 by Justincc (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
[10:59]  Connecting to in-world Voice Chat...
[10:59]  Connected
[10:59]  Nebadon Izumi: hello
[10:59]  BlueWall.Slade Hey
[10:59]  Seth.Nygard is Online
[10:59]  Aine.Caoimhe hi Neb
[10:59]  Justin Clark-Casey: hi neb, folks
[11:00]  Aine.Caoimhe hi Justin
[11:00]  BlueWall.Slade Hi
[11:00]  Justin Clark-Casey: bluewall: btw, your commit 12108bf had tabs in it
[11:00]  Seth.Nygard waves to everyone
[11:00]  BlueWall.Slade Uhoh
[11:00]  BlueWall.Slade I need to set my IDE
[11:00]  Justin Clark-Casey: :)
[11:01]  Aine.Caoimhe crucification for him, or just a good flogging?
[11:01]  BlueWall.Slade I did notice some and changed the to spaces, but I guess I missed some.
[11:01]  Justin Clark-Casey: I would except for the fact I'm sure I'll do it myself some time :)
[11:01]  Justin Clark-Casey: I prefer to avoid the floggings
[11:01]  Dahlia.Trimble is Online
[11:01]  BlueWall.Slade I have new software here & some things are still not 100% setup yet.
[11:01]  Justin Clark-Casey: bluewall: yeah, I look at your next commit and it didn't touch all of them - didn't knowthat was the intnetion
[11:02]  Alicia.Raven hi
[11:02]  Justin Clark-Casey: hi alicia
[11:02]  BlueWall.Slade I'll go in there an clean the rest up
[11:02]  BlueWall.Slade Hi
[11:03]  Andrew Hellershanks is Online
[11:03]  Richardus.Raymaker somebody here that worked with KFM ?
[11:03]  Richardus.Raymaker or Key Frame Motion
[11:03]  BlueWall.Slade Yes
[11:03]  Nebadon Izumi: I have never used Kerframes myself yet
[11:04]  Richardus.Raymaker i have weird problem. it works fine. as soon i start to add the rotation. it start to spin.
[11:04]  Richardus.Raymaker i used it already many times. but sometimes it start to spin
[11:04]  BlueWall.Slade hmm
[11:04]  Richardus.Raymaker checked the KFM list but the rotation data looks fine
[11:05]  BlueWall.Slade you know the rotation is just the delta between the current reotation and the target rotation?
[11:05]  Richardus.Raymaker hmm. thats maby why the divide where there. aha. need to check that
[11:06]  Richardus.Raymaker ok. i think thats the problem
[11:06]  Richardus.Raymaker need to check that
[11:06]  BlueWall.Slade Each step is just the delta to the next - postition too
[11:06]  Richardus.Raymaker i know with position. but my mind did not triggered with that on rotation :O
[11:06]  Richardus.Raymaker almost finaly have a good curve for my train.
[11:06]  BlueWall.Slade :)
[11:07]  Dahlia.Trimble is Online
[11:07]  Dahlia.Trimble hi
[11:07]  BlueWall.Slade Hello
[11:07]  Richardus.Raymaker hi dahlia
[11:07]  Nebadon Izumi: hello Dahlia
[11:08]  Aine.Caoimhe I've begun investigating an odd issue that Nara mentioned to me and I was finally able to seems that if an NPC is wearing a scripted attachment there are occasions where later removing the NPC somehow still manages to tie up a script thread which then never gets released (until sim is restarted). If you do this often enough it can cause all scripts in the sim to cease responding or become extremely unresponsive. Is it possible that it's trying to save script state or something like that? Or perhaps if a thread is currently in use by the scipt at the time it is removed from the scene the thread isn't released?
[11:08]  Key Gruin: hi folks
[11:08]  Richardus.Raymaker is opensim someday getting some CDN (content delivery network) ? that way connections at home need to send less data because assets can be placed soemwhere else that are used on the simulator ?
[11:08]  BlueWall.Slade Hi
[11:09]  Richardus.Raymaker hi key
[11:09]  Kayaker.Magic How come: whenever I TP to this grid, all the scripts in my suitcase and attachment say "the script is not in the database"
[11:09]  Nebadon Izumi: we block incoming scripts
[11:09]  Nebadon Izumi: prevent greifing etc..
[11:10]  Dahlia.Trimble hi Key
[11:10]  Justin Clark-Casey: aine: I have not heard of that one before
[11:10]  Kayaker.Magic But all I had to do was paste them in again. I had a booth and permission to rez there. Scripts worked fine if I carried them in notecards.
[11:10]  Nebadon Izumi: technically we can pretty much already do that Richardus
[11:10]  Aine.Caoimhe nor had I until Nara mentioned it, but then I was able to reproduce it with some effort
[11:10]  Nebadon Izumi: we infact do it here sort of, using caps we tell the viewer to fetech textures from the robust server
[11:10]  Nebadon Izumi: and not the region
[11:10]  Richardus.Raymaker nice
[11:11]  Nebadon Izumi: its not a default behavior of opensimulator, but it can be done
[11:11]  Nebadon Izumi: just by changing configurations a bit
[11:11]  Justin Clark-Casey: aine: it's the kind of thing that might possible be fixed with co-op script stopping
[11:11]  Justin Clark-Casey: aine: unless she has that set already. It's curretnly only default in dev code
[11:11]  Aine.Caoimhe I ended up rezzing an array of NPCs wearing scripts and then removed them....rinse and repeat for a while and eventually the region hung
[11:12]  Aine.Caoimhe I tested it in my place which is running coop
[11:12]  Aine.Caoimhe and on a pimple island in an empty sim so nothing else was doing anything
[11:12]  Justin Clark-Casey: aine: ok, will need a bug report on that. Iknow there's not a lot of activity there atm but this is an issue I would want to looka t before release
[11:12]  TP Tone Thunder: Script running
[11:12]  Dahlia.Trimble how to enable coop?
[11:12]  Aine.Caoimhe it took me about 200 rez & kill before it eventually happened
[11:13]  Key Gruin: in OpenSim.ini Dahlia
[11:13]  Aine.Caoimhe it's a setting in XEngine
[11:13]  Justin Clark-Casey: ScriptStopStrategy = co-op in [XEngine]. It should trigger recompile of all scripts automatically but there are some bugs I also need to investigate
[11:13]  Dahlia.Trimble whats the setting?
[11:13]  Richardus.Raymaker Murder ! :)
[11:13]  Key Gruin: true or false? I think
[11:13]  Dahlia.Trimble lol the name
[11:13]  Aine.Caoimhe no, default is (or was) abort
[11:13]  Justin Clark-Casey: ScriptStopStrategy = co-op
[11:13]  Key Gruin: or just specifying it
[11:13]  Dahlia.Trimble never mind I'll look for it
[11:13]  Key Gruin: ya
[11:13]  Aine.Caoimhe so you change from abort to co-op
[11:14]  Dahlia.Trimble ty justin
[11:14]  Aine.Caoimhe I'll write a simplified script to reproduce it
[11:14]  Justin Clark-Casey: thanks aine.
[11:14]  Aine.Caoimhe and I'll also try setting that sim with far fewer script threads available and see if that speeds it up
[11:14]  Justin Clark-Casey: I'm stil working on the ghosts merge. Quite close but I want to think about a few final things
[11:15]  Dahlia.Trimble does coop interact with AppDomain?
[11:15]  Justin Clark-Casey: also, I want to question Mic about the adaptive throttles stuff
[11:15]  Richardus.Raymaker hi justin
[11:15]  Justin Clark-Casey: dahlia: no, it should be orthagonal
[11:15]  Dahlia.Trimble kk
[11:15]  Justin Clark-Casey: the adaptive throttles stuff is modelled on TCP congestion avoidance algorithms but there are some major differences
[11:15]  Justin Clark-Casey: which might make it much harsher than it is with tcp
[11:16]  Aine.Caoimhe I've also encountered another odd bug -- it's one I think has been happening for a while but has taken a long time to pin relates to making an HG tp to a simulator that is hosted on the same computer as the viewer but is connected to a grid hosted outside that LAN
[11:16]  Dahlia.Trimble from the way you've described it justin it sounds way too aggressive
[11:17]  Aine.Caoimhe at some point during the tp home process, it switches from my external facing IP to my internal network IP and then fails the tp due to mismatch
[11:17]  Justin Clark-Casey: dahlia: yeah, Ithink so
[11:18]  Dahlia.Trimble cutting it in half for a missing packet when you dont even know why the packet was lost. Could have been a temporary internet glitch
[11:18]  BlueWall.Slade Aine, The external grid sees your external ip address, and when you jump to your local region, it sees your private ip address.
[11:18]  Dahlia.Trimble should be some averaging done or something
[11:18]  BlueWall.Slade The do not mathc, so it fails
[11:18]  BlueWall.Slade match.
[11:18]  Aine.Caoimhe right
[11:18]  Aine.Caoimhe but it shouldn't be doing that
[11:18]  BlueWall.Slade it always has.
[11:19]  Justin Clark-Casey: dahlia: tcp is the same but I believe it doesn't do lots of halving at once
[11:19]  Justin Clark-Casey: which may be the issue, where a whole string of packets expire in quick succssion
[11:19]  Aine.Caoimhe so basically HG tp to any region you host yourself = intended fail?
[11:19]  Dahlia.Trimble tcp will recover
[11:19]  Aine.Caoimhe (unless you also host the grid yourself)
[11:20]  BlueWall.Slade aine, I guess you could do some routing to make your local region think it came from your external network.
[11:20]  BlueWall.Slade Hg thinks the ip is being forged.
[11:21]  Justin Clark-Casey: dahlia: yeah that's another issue, our recovery is not as quick as tcp
[11:21]  Dahlia.Trimble have you tried just increasing ack timeout?
[11:22]  Justin Clark-Casey: it's already 60 seconds. At that point I expect the packets will never be received
[11:22]  Justin Clark-Casey: or am I wrong
[11:22]  Dahlia.Trimble 60?
[11:22]  Dahlia.Trimble heh
[11:22]  Justin Clark-Casey: no, that's rubbish, sorry thinking of something else
[11:22]  Justin Clark-Casey: But I do believe the packets are permanently lost and it's not just late delivery
[11:22]  Richardus.Raymaker 60 sounds very long
[11:24]  Justin Clark-Casey: it's one of those things that may not normally happen but we saw problems even under faiirly moderate load on the conf grid
[11:24]  Dahlia.Trimble what kind of problems?
[11:24]  Justin Clark-Casey: strings of packets expiring and forcing connections down to extremely low throttles
[11:25]  Aine.Caoimhe I guess as a "design" it seems odd that if you connect a region to an external grid you lose the ability to TP from an HG location back to your own regions without a lot of (above my skill level) network hacking and can only TP back to some other region hosted by your home grid first
[11:25]  BlueWall.Slade Aine, ATM, yes.
[11:25]  Richardus.Raymaker But why does TP hom e works on HG ?
[11:25]  Dahlia.Trimble ya I can see it forcing them down
[11:25]  Nebadon Izumi: ya i experienced it, it was wierd, i had 2 viewers running on same computer
[11:26]  Nebadon Izumi: 1 avatar was throttled other not
[11:26]  Dahlia.Trimble but what happens if a few packets are lost in a row? is it halved several times?
[11:26]  BlueWall.Slade Aine, ask around though. There may be a workaround. That is the wy it works here.
[11:26]  Aine.Caoimhe @Richardus it works fine if you're also hosting your own grid on the same LAN
[11:26]  Justin Clark-Casey: dahlia: yes, that's the major issue
[11:27]  Dahlia.Trimble adaptive stuff is difficult but I can't see it working at al lthe way you described it
[11:28]  Justin Clark-Casey: dahlia: the adaptive has been default for 2+ years so somehow it has
[11:28]  Nebadon Izumi: it seemed like once it started to throttle it wouldnt stop
[11:28]  Justin Clark-Casey: but under stress it might show up the problems
[11:28]  Nebadon Izumi: even if i relog avatar it was instantly throttled again
[11:28]  Dahlia.Trimble are you certain the packes were sent in the first place?
[11:28]  Justin Clark-Casey: it might be that under low loads packets are lost extremely seldom
[11:28]  Nebadon Izumi: ya for me part of my problem was having 40-85 avatrs logged in from same connection
[11:28]  Justin Clark-Casey: the expiry is when the server receives no ack for a reliable packet
[11:28]  Nebadon Izumi: i was basically like a noise generator
[11:28]  Nebadon Izumi: hehe
[11:28]  Dahlia.Trimble ya I never see packet loss in the viewer stats window
[11:29]  Justin Clark-Casey: within a certain teimframe
[11:29]  Dahlia.Trimble I wonder if under load there may be some other failure and it doesnt even send the packet
[11:29]  Justin Clark-Casey: tbh, nebadon might be an extreme case but others experience it too, including myself on one occasion
[11:29]  Justin Clark-Casey: dahlia: I don't think so
[11:29]  Justin Clark-Casey: but who knows
[11:30]  Justin Clark-Casey: in any case, moving to manual throttling resolved the issues
[11:30]  Nebadon Izumi: ya normally i didnt experience it, it wasnt until i degraded my network so much that i couldnt really do much of anything, couldnt even load at times
[11:30]  Nebadon Izumi: that i experienced these things
[11:30]  Justin Clark-Casey: but that means the server no longer throttles back in response to a genuinely bad connection, or one where somebody has set a bw viewer setting higher than it can cope with
[11:31]  Dahlia.Trimble well it could average packet oss
[11:31]  Dahlia.Trimble *loss
[11:31]  Justin Clark-Casey: I think one would try and keep the algorithm as simple as possible
[11:31]  Nebadon Izumi: what was odd, was my local avtar would get throttled, but my HG one never did
[11:31]  Dahlia.Trimble one packet could be anything and halving it is extreme
[11:31]  Nebadon Izumi: why it would always choose the local over the HG confuses me
[11:32]  Aine.Caoimhe and then a thrid issue relating to HG attachments....when you tp into a region that is fairly busy there are occasions where the region seems to lose track of what it has sent and what it hasn't and other existing agents in the region never get the data. Relogging doesn't solve this....the only way the others will ever see the attachment is if *they* relog
[11:32]  Nebadon Izumi: probably comes back to packet loss Aine
[11:32]  Nebadon Izumi: they never get the packets first time, only solution is relog
[11:32]  Dahlia.Trimble that sounds more like an asset timeout
[11:33]  Aine.Caoimhe so some sort of "force update attachments" would be king of handy
[11:33]  Nebadon Izumi: don't we have that?
[11:33]  Aine.Caoimhe like a rebake
[11:33]  Nebadon Izumi: or is that rebake only?
[11:33]  Nebadon Izumi: ok ya rebake only i guess
[11:33]  Aine.Caoimhe no, force update only updates scene objects but not attachements
[11:33]  Aine.Caoimhe and rebake only updates wearables
[11:33]  Aine.Caoimhe but neither updates attachments
[11:34]  Nebadon Izumi: i wonder if you right click your folder for your avatar and replace appearance
[11:34]  Nebadon Izumi: if it would blast out updates again
[11:34]  Seth.Nygard Ideally I would think a rebake should also update attachments
[11:34]  Nebadon Izumi: probably would
[11:34]  Nebadon Izumi: save a relog
[11:34]  Aine.Caoimhe even if I relog it doesn't update for anyone who is still in the region
[11:34]  Aine.Caoimhe the only way they see the attachment is if *they* relog
[11:35]  BlueWall.Slade Is that something to do with the xbakes service?
[11:35]  BlueWall.Slade I dont' know if attachements are included in that -0I doubt it.
[11:36]  Aine.Caoimhe for people who wear mesh clothing it's kind to know that eventually everyone else will stop seeing you as stark naked without their having to relog
[11:36]  Justin Clark-Casey: bluewall: not related
[11:36]  Justin Clark-Casey: bakes only relate to the avatar textures, not it's attached scene objects
[11:37]  BlueWall.Slade Ok, thanks
[11:37]  BlueWall.Slade So, should the forced reload affect the attachments?
[11:38]  Justin Clark-Casey: forced reload?
[11:38]  Aine.Caoimhe ideally when an avi rebakes it would also trigger a "resend attachments" as part of that
[11:38]  BlueWall.Slade sometning like that - they were talking about it a cpl min agao
[11:38]  Justin Clark-Casey: aine: rebakes and attachemnts are conceptually seperate
[11:38]  Aine.Caoimhe kk
[11:38]  BlueWall.Slade [11:33] Aine.Caoimhe no, force update only updates scene objects but not attachements
[11:39]  Justin Clark-Casey: indeed, rebakes can't even be triggered from the console except under some circumtsances due to the way the viewr
[11:39]  Justin Clark-Casey: works
[11:39]  Justin Clark-Casey: bluewall: ok. THe force update command is ancient. I think one could justify making it recsend attachments too tbh
[11:39]  BlueWall.Slade ++
[11:39]  Justin Clark-Casey: although that will end up being a lot of data for a reasonably populated region
[11:40]  BlueWall.Slade And ideas about what might make it miss them to start with?
[11:40]  Aine.Caoimhe can't always rely on someone having console access to force it from the server side even if the command existed so it would be nice for the avatar wearing the stuff to be able to trigger other people seeing it
[11:40]  Justin Clark-Casey: something like "attachments send" might be better - another command in the AttachmentsCommandModule
[11:40]  Justin Clark-Casey: aine: right, this would only really be useful for emergencies and debugging
[11:41]  BlueWall.Slade Is it anything to do with the Teleports, the order in which things are moved and connection speeds?
[11:41]  Aine.Caoimhe the interesting thing (which might defeat that) is that even relogging yourself doesn't make that happen
[11:42]  Justin Clark-Casey: aine: that is curious
[11:42]  Aine.Caoimhe yeah
[11:42]  Justin Clark-Casey: aine: what code level is this?
[11:42]  Aine.Caoimhe I had it happen in the LBSA HG regioon
[11:43]  Aine.Caoimhe half the people there on Saturday when I arrived got my body mesh but not my dress
[11:43]  Aine.Caoimhe I relogged a few times to no avail
[11:43]  Nebadon Izumi: yea i had similar problems
[11:43]  Aine.Caoimhe but when anyone who was seeing me naked later relogged they then saw me perfectly
[11:43]  Nebadon Izumi: i find that standalone definitely doesnt run as well as grid mode
[11:43]  Justin Clark-Casey: I would like to see what beahviour is once I've merged code
[11:43]  Nebadon Izumi: which is not all that surprising
[11:44]  Key Gruin: speaking of standalone mode, WRight Plaza is now up as a hypergrid standalone
[11:44]  Aine.Caoimhe this was happening between the RG server and LBSA server which are both pretty high speed I assume
[11:45]  Sarah Kline: this is mainly affecting mesh attachments or all
[11:45]  Aine.Caoimhe it affects all
[11:45]  Key Gruin: if folks want to hop over there after the meet
[11:45]  Nebadon Izumi: ya speed of hardware and network doesnt factor in much
[11:45]  Aine.Caoimhe but more likely to occur with mesh
[11:45]  Aine.Caoimhe (or seems to be)
[11:45]  Nebadon Izumi: the more you pile onto a single mono instance the worse performance will get
[11:45]  Nebadon Izumi: same reason why running 2 regions in a single simulator is bad
[11:45]  Key Gruin: Plaza-HG
[11:46]  Andrew Hellershanks: I have some grid owners who feel that 0.8 isn't as good as 0.7.6 as 0.8 seems to be laggy compared to 076.
[11:46]  Andrew Hellershanks: If we want to start discussing performance.
[11:46]  Dahlia.Trimble oh the real wright is back up?
[11:46]  Nebadon Izumi: seemes is hardly scientific though
[11:46]  Nebadon Izumi: Dahlia key loaded an OAR into standalone
[11:47]  Nebadon Izumi: on the same server as the real wright plaza
[11:47]  Key Gruin: ya it's the latest Wright before the downtime
[11:47]  Nebadon Izumi: we can jump over and try
[11:47]  Marcus.Llewellyn Seeing as 0.8 as largely a product of the testing that went into OSCC13, and that OSCC13 went so well, I'm not inclined to believe 0.7.6 is a better performer than 0.8.
[11:47]  Nebadon Izumi: i suspect it may not be a great experience, but you never know could be surprised
[11:48]  Nebadon Izumi: Wright Plaza is a nasty beast for a standalone none of us have ever visited
[11:48]  Dahlia.Trimble I had an older wright up but it crashes every few days
[11:48]  Key Gruin: could surely use some tweaking on the configs
[11:48]  Nebadon Izumi: Performance is a subjective thing
[11:48]  Marcus.Llewellyn It's not using SQlite, is is?
[11:48]  Nebadon Izumi: peoples feelings need to not factor into it
[11:48]  Nebadon Izumi: heh
[11:48]  Sarah Kline: lol
[11:48]  Key Gruin: seems to be running good with no visitors :P
[11:48]  Nebadon Izumi: hard data is good
[11:48]  Dahlia.Trimble lol
[11:48]  Richardus.Raymaker Well, what hardware are the people useing that say 0.8 is slower. it not feels slower for me. maby assets load a fraction slower. i think more depends onm memory
[11:48]  Key Gruin: lol Marcus no
[11:48]  Marcus.Llewellyn Phew. :)
[11:49]  Key Gruin: :)
[11:49]  Justin Clark-Casey: asying it is "slower" means nothing. ONe would need actual numbers.
[11:49]  Richardus.Raymaker only wish that scripting where faster
[11:49]  Andrew Hellershanks: Nebadon, true. They used to promote the grid by telling people they won't experience the lag that is so typical of SL and that used to be true but it isn't holding up any more as of 0.8
[11:50]  Aine.Caoimhe wishes there was an ini setting for Allow_llSleep = false
[11:50]  Nebadon Izumi: yea performance will decrease as you attract more customers
[11:50]  Sarah Kline: and add more content
[11:50]  BlueWall.Slade The 100 region region?
[11:50]  Nebadon Izumi: opensimulator doesnt scale to extremes like SL without a lot of investment in the technology
[11:50]  BlueWall.Slade Var region
[11:50]  Joe.Radik Is there anything like a performance test suite available?
[11:50]  Key Gruin: that's the main diff between SL and OS is MONEY
[11:50]  Marcus.Llewellyn I think what most people mean when they say that an old version is "better" is really that it's more familiar. It has bugs and quirks they know and are used to working around. A new version may truely be better in many ways, but it will also have new quirks, resultsing in complaints of, "my scripts are different," or, "my avatar moves funny now."
[11:51]  Justin Clark-Casey: right, it's like saying you should be able to take an Apache instance and use it to run Facebook
[11:51]  Nebadon Izumi: right
[11:51]  Andrew Hellershanks: The biggest pain right now is crashing (almost) every time you TP from a var to a regular region.
[11:51]  Nebadon Izumi: so much more goes into getting there than just running apache
[11:51]  Justin Clark-Casey: joe: Not really - the nearest is probably to use the stats recording on comparative loads
[11:51]  Richardus.Raymaker its bad to alllow llSleep. , yopu need it but dont make times higher then 1 sec
[11:51]  BlueWall.Slade I dont' see that here Andrew
[11:51]  Andrew Hellershanks: If it doesn't crash, the viewer appears to lock up
[11:51]  Aine.Caoimhe my avatar *does* move funny can't navigate tiny steps :p
[11:51]  Dahlia.Trimble if they notice a difference then the older becomes better
[11:52]  BlueWall.Slade I Hg between a var region and normal region , jump around between normal and var on my grid.
[11:52]  Andrew Hellershanks: There is a version in git master where it doesn't seem to happen any more but that version was a merge so no idea what changes to back port to 08 to fix the TP crashing.
[11:52]  Marcus.Llewellyn +1 Dahlia for succinctness. :)
[11:52]  Andrew Hellershanks: Blue, what version of code are you using?
[11:52]  Dahlia.Trimble but I guess you cant blame them for complaining if something they were relying on stops working
[11:53]  BlueWall.Slade Is this code pulled straight from the repository, or does it have modules andproprietary code added?
[11:53]  BlueWall.Slade Master gid
[11:53]  BlueWall.Slade git.
[11:53]  BlueWall.Slade I have some modules and a small number of changes
[11:53]  Nebadon Izumi: I don't mind people complaining, as long as that isnt all they are doing
[11:53]  Justin Clark-Casey: I msut admit, if you add modules, let along proprietary stuff, things become very difficult
[11:53]  BlueWall.Slade nothing performance related, just features.
[11:53]  Nebadon Izumi: they need to get more involved with data and willingness to try to recreate the circumstances
[11:53]  Justin Clark-Casey: it would be very easy to cause problems by even mildly non-optimal region module code
[11:54]  Justin Clark-Casey: neb: it's easy just to complain :)
[11:54]  Justin Clark-Casey: easier
[11:54]  Dahlia.Trimble I have added proprietary modules, they never cause any problems with master
[11:54]  Andrew Hellershanks: Nebadon, I've sometimes complained about stuff and if I've done it long enough someone digs in to the issue and finds there was a problem that needed fixing. :)
[11:54]  Nebadon Izumi: im the complainer in cheif so
[11:54]  Dahlia.Trimble but sometimes master causes problems with them :/
[11:54]  Key Gruin: hehe
[11:54]  Aine.Caoimhe whines constantly....just ask Justin :p
[11:54]  Dahlia.Trimble usually an API change
[11:55]  Andrew Hellershanks: I'm happy to help with testing changes but the issues that are being reported to me are beyond my ability to fix in OS.
[11:55]  BlueWall.Slade ++ Dahlia
[11:55]  Justin Clark-Casey: there is no API
[11:55]  Marcus.Llewellyn There is only Zuul.
[11:55]  BlueWall.Slade But, those are usually easy to fix.
[11:55]  BlueWall.Slade Haha JCC ++++++++
[11:55]  Dahlia.Trimble justin lol
[11:55]  Dahlia.Trimble there is an api, its just totally undocumented :P
[11:56]  Dahlia.Trimble and very unstable
[11:56]  Justin Clark-Casey: which makes it not an api
[11:56]  Andrew Hellershanks: AH, got a confirmation that lag is mostly seen in vars so I'm not as concerned about that.
[11:56]  Billy.Bradshaw Some people do fiddle with the version number, so when you visit you see A1 Special or something, so no idea what version is actually being used
[11:56]  Andrew Hellershanks: The 10x10 is gone It wasn't useable. Max being used is now about 5x5
[11:56]  BlueWall.Slade Andrew - how bif ar the vars? and content?
[11:56]  BlueWall.Slade ok
[11:56]  Dahlia.Trimble well then stuff like "IClientAPI" is not properly named
[11:56]  Justin Clark-Casey: yes, it is misnamed
[11:57]  BlueWall.Slade That would be about the same as running a 5 regions in a standalone
[11:57]  Andrew Hellershanks: Blue, I haven't checked lately. I don't get in to the content that much. Others do the building of what is in regins.
[11:57]  Dahlia.Trimble and LSL_Api
[11:57]  Nebadon Izumi: Andrew 10x10 is not good idea, the ultimate max you should never go beyond 2048x2048
[11:57]  Justin Clark-Casey: LSL is different, that actually is a documented thing
[11:57]  Nebadon Izumi: but even that is pushing it, 1024x1024 var really should be the upper limit right now
[11:57]  Justin Clark-Casey: that comes back to the question of putting sensible limits in code
[11:57]  BlueWall.Slade I think kVar regions are great for creating spacious environments. But, I think that you want to not load them up to heavily.
[11:57]  Nebadon Izumi: we shouldnt officially support anything larger than that at this time
[11:57]  Justin Clark-Casey: rather than allowing peopel to think they can run huge var regions by default
[11:58]  Justin Clark-Casey: like the prim limit
[11:58]  Dahlia.Trimble libomv used to change API at the whim of a hurlicane and so many people complained that they finally made an effort to keep it consistent
[11:58]  Nebadon Izumi: well im not saying hard code it, but there needs to be a big sign warning you about the dragons that will ruin your fun
[11:58]  Aine.Caoimhe hehe
[11:58]  Andrew Hellershanks: Nebadon, the people running the grid felt you said 10x10 would be ok which is why they tried it. :)
[11:58]  Justin Clark-Casey: nebadon: not hardcode, but make it so you have to change config to do something different
[11:58]  Nebadon Izumi: again with the feelings!
[11:58]  Nebadon Izumi: lol
[11:58]  Nebadon Izumi: feelings get you in trouble :P
[11:59]  Andrew Hellershanks: I thnk they were going on old info.
[11:59]  Nebadon Izumi: DATA!!!
[11:59]  Richardus.Raymaker sofar a 3x3 looks fine on var. bigger is load to slow
[11:59]  Aine.Caoimhe do your best Joe Friday impression
[11:59]  Nebadon Izumi: ya im just kidding, its fun to experiment
[11:59]  Nebadon Izumi: as long as you keep your expectations low
[11:59]  Andrew Hellershanks: ok, maybe not felt but recalled you saying that but I did try and dissuade them.
[11:59]  Nebadon Izumi: :D
[11:59]  BlueWall.Slade Andrew - people should always take baby steps to implement their ideas from a known good working environment
[12:00]  Nebadon Izumi: hey i am all for the lesson in pain
[12:00]  Andrew Hellershanks: yeah. Only way to know what will happen is to try something.
[12:00]  Justin Clark-Casey: if we were dealing with such ppl we probably woudln't; be having this conversation/....
[12:00]  Nebadon Izumi: sometimes it takes the painful lessen to get on board
[12:00]  BlueWall.Slade Test, Test, Test, Document, Debug ...
[12:00]  Key Gruin: some people like to dive in headfirst
[12:00]  Joe.Radik ,,,take a step back...
[12:00]  Richardus.Raymaker its not Test Test Test SCREAM Test Tes etc. ? :)
[12:00]  Billy.Bradshaw I have very large vars, but these are used as geo projects where there are virtually no prims, but supports upwards of 10 avis so far.
[12:01]  Richardus.Raymaker i really wish opensim did have good plants
[12:01]  Nebadon Izumi: ya the big problem is the viewer mostly
[12:01]  Andrew Hellershanks: Blue, True. Makes sense to those of us involved in coding but it can e harder to convince something that some new great feature may not be fully ready to be jumped on and used.
[12:01]  Nebadon Izumi: with the terrain, a 10x10 LL terrain eats up a lot of memory
[12:01]  Nebadon Izumi: possibly into the Gigabyte range
[12:01]  Andrew Hellershanks: Richardus, Plants as in ones for a garden?
[12:01]  BlueWall.Slade I think it is a misconception to think that a var region should support multiples of the 256x256 avatar/content load depending on their size.
[12:02]  Billy.Bradshaw Test take data and adjust your expectations accordingly
[12:02]  Dahlia.Trimble someone could write a procedural plant module :)
[12:02]  Seth.Nygard waits for someone to try a busy club in a 2048x0248 or bigger var region full of objects and wonder why it performs so poorly.
[12:03]  Richardus.Raymaker OMG
[12:03]  Nebadon Izumi: you guys want to try a jump over to Wright Plaza stadnalone before we head out?
[12:03]  Aine.Caoimhe "busy club in a var" is a contradiction in terms....too many people still running 0.7.6 so they can't tp there to make it busy
[12:03]  Dahlia.Trimble sure
[12:03]  Nebadon Izumi: what is the address again Key?
[12:03]  Marcus.Llewellyn Yay Wright Jump 2014!
[12:03]  Andrew Hellershanks: what can I do about the var to regular TP? The crashing is a real pain
[12:03]  Richardus.Raymaker 10x10 is arounbd 3GB i guess. or more 5x5 where already 2.5GB
[12:03]  Seth.Nygard Load test in Wright Plaza then ...
[12:03]  Key Gruin: Plaza-HG
[12:03]  Richardus.Raymaker empty
[12:03]  Nebadon Izumi: Andrew that should only happen on older versions i thought
[12:04]  Dahlia.Trimble hops... laterz all :)
[12:04]  Justin Clark-Casey: andrew: not much except report it and I may eventually look at it., Though as var regions are experimental it's not going tob e amust fix
[12:04]  Andrew Hellershanks: It happens in latest 08PF
[12:04]  Justin Clark-Casey: bye dahlia
[12:04]  Nebadon Izumi: ive not seen that
[12:04]  Justin Clark-Casey: or perhaps misterblue will
[12:04]  Justin Clark-Casey: could be a viewer issue anyway
[12:04]  Nebadon Izumi: i jump between Var and single size regions all the time
[12:04]  Key Gruin: see ya over at Wright Plaza-HG :)
[12:04]  Nebadon Izumi whispers: im jumping over, see you there :)
[12:04]  Nebadon Izumi: Plaza-HG
[12:04]  Andrew Hellershanks: Neb, in what version of code?
[12:04]  Justin Clark-Casey: hang on, what's the address@
[12:04]  Justin Clark-Casey: ok, great
[12:04]  Key Gruin: Plaza-HG
[12:05]  Region found!
Personal tools
About This Wiki