Chat log from the meeting on 2018-02-06

From OpenSimulator

Jump to: navigation, search

[10:58] Leighton.Marjoram can you both see me ok my outfit is a little high complexity wise
[10:59] Sheera Khan: I see you only partially ...
[11:00] Sheera Khan: aahhhhhh there you go Leighton :-)
[11:00] Leighton.Marjoram oh my its worse than I thought
[11:00] Kayaker Magic: I still see a big transparancy through your middle Leighton
[11:01] Leighton.Marjoram that a lovely mesh jacket but my complexity is something like 191000
[11:01] Selby.Evans I saw that transparency, but now it is filled in
[11:02] Selby.Evans Just took a long time to load the texture
[11:02] Sheera Khan: now your shoes are here as well Leighton :-)
[11:02] Kayaker Magic: What is it about this place that makes mesh load so slow?
[11:03] George Equus: waiting for something to sit on  :)
[11:03] George Equus: Extremly sloooow
[11:03] Leighton.Marjoram nods @ Kayaker
[11:03] Sheera Khan: I don't know either.. I think this region is extremly slow
[11:03] George Equus: Always been from the start, for me
[11:04] Leighton.Marjoram it is strange I dont see slow mesh load times elsewhere
[11:04] George Equus: Wild guess... I sit at least
[11:04] Sheera Khan: these benches really should be in the cache by now... but nooooo, they take ages to load
[11:05] George Equus: loaded but no cushions yet
[11:05] Kayaker Magic: The cushions came in for me instantly today, only the backs were slow.
[11:05] Andrew.Hellershanks Better that it is no cushions rather than no clothes. :)
[11:06] George Equus: Andrew floating in mid air no chairs in sight so far
[11:06] Kayaker Magic: And Leighton has finally rezzed for me!
[11:06] Andrew.Hellershanks I thought I was on the wrong place at first but realized I came in to this region over by the other building.
[11:06] Sheera Khan: after only 5 minutes ;-)
[11:06] Leighton.Marjoram wow thats a long while
[11:06] George Equus: my clothes res pretty instantly on any other region
[11:06] Andrew.Hellershanks The chairs in this corner do take a long time to rez properly.
[11:07] Ubit Umarov: i was here bf.. so i did see the chairs etc when i arrived :p
[11:07] Andrew.Hellershanks So much for finding a region without lag for these meetings. :P
[11:07] George Equus: Maybe Wright wasn't that bad after all....
[11:07] Andrew.Hellershanks I should ask nebadon to rez one of the stools from the old meeting location.
[11:08] Andrew.Hellershanks George, ask Ubit. He was the one who had a lot of problems with the old location.
[11:08] Ubit Umarov: forget wright.. that was the worse possible location
[11:08] Kayaker Magic: I have some strange complaints about the latest version of FireStorm and its interaction with OpenSim:
[11:08] Ubit Umarov: 35K prims really ??
[11:08] George Equus: I know Andrew.. and I never saw any lag over there ever practically
[11:09] Kayaker Magic: On OpenSim the new FireStorm does not display newly imported mesh.
[11:09] Andrew.Hellershanks George, I rarely saw lag issues over there.
[11:09] Sheera Khan: here we have less than 900 and it is not much better ^^
[11:09] Andrew.Hellershanks Kayaker, which version of Firestorm are you using?
[11:09] Kayaker Magic: But OpenSim 0.9.1 built from the sources, FireStorm can display mesh fine.
[11:09] Sheera Khan: it does display all items here though
[11:10] George Equus: I did take the plunge... and updated FS to latest. So far seem fine
[11:10] Sheera Khan: I'm on 5.0.11.OS 64Bit
[11:10] Kayaker Magic: What has happened between 0.9.0 and 0.9.1 that would effect how FS displays mesh?
[11:10] Leighton.Marjoram me too George been about a week and no issues
[11:10] Sunshine Szavanna: brb some things are not rezzing
[11:10] Ubit Umarov: nothing..
[11:10] Andrew.Hellershanks Sheera, that's the one I'm using.
[11:10] Sheera Khan: works fine for me
[11:10] Ubit Umarov: FS is just not showing meshs it did upload badly i think
[11:11] Sheera Khan: that was more of an error user side...
[11:11] Ubit Umarov: a few cases 0,9.1 does refuse to upload also
[11:11] George Equus: FSEnforceStrictObjectCheck need to be FALSE though... or my trousers disappear  :)
[11:11] Sheera Khan: if you use the analyze Button for the upload you're in for a surprise afterwards
[11:11] Kayaker Magic: Ahh! We can't have that!
[11:11] Ubit Umarov: some clever ppl told 0,9.1 was broken doing that and told ppl to remove that code.. well whatever
[11:12] Ubit Umarov: it is not analyse per si.. comments are wrong on that
[11:12] Sheera Khan: George, I see your trousers just fine I think
[11:12] Ubit Umarov: just analyse is more complex.. viewers fail more and don't detect it
[11:12] Sheera Khan: and that setting is default here
[11:13] Ubit Umarov: so yeap not that recommended for now..
[11:13] George Equus: analyze apparently mess up thing badly unless on SL...
[11:13] Ubit Umarov: but does have a few merits, specially for bullet.. well if bullet did work :p
[11:13] Andrew.Hellershanks hm... I just found a bug in FS.
[11:13] George Equus: FS team make a special note NOT to use that on Opensim
[11:13] Ubit Umarov: it also messes at sl
[11:14] Ubit Umarov: that FS code is also for sl
[11:14] Ubit Umarov: just havoc mesh tools fail a bit less
[11:14] Andrew.Hellershanks The notes in the wiki about uploading mesh also states that analyze mesh should not be used.
[11:14] George Equus: precisely
[11:14] Ubit Umarov: ( but if you go sl with fs-OS it does not have havoc mesh tools
[11:15] George Equus: no
[11:15] Ubit Umarov: and it sais what i just said.. not recomended (for now )
[11:15] George Equus: still reading the release notes on the viewer
[11:15] George Equus: only couple of days old so time will tell I guess
[11:16] Andrew.Hellershanks Kayaker, are you using FS 5.0.11?
[11:16] George Equus: I do like the info you get now on Inventory items
[11:16] Andrew.Hellershanks George, a couple of days? What is the current version number?
[11:17] Kayaker Magic: No, I'm back on 5.0.1 for now.
[11:17] George Equus: 5.0.11
[11:17] Ubit Umarov: the physics mesh view per prim is also nice
[11:17] Ubit Umarov: well will not match physics unless on ubOde the the reasons well known
[11:18] Ubit Umarov: hmm will not match on some cases..
[11:18] George Equus: Released around Jan 20th I think
[11:19] George Equus: Time flies...
[11:19] Ubit Umarov: for some odd reason does not work on normal prims or sculpts only mesh
[11:19] Ubit Umarov: convex shape specially
[11:20] Ubit Umarov: but the full screen display of physics shapes also did not work on those before
[11:20] Andrew.Hellershanks Nuts. I've locked up Chrome again. :P I was going to check something in the wiki. I'll have to do that later.
[11:22] Andrew.Hellershanks There is one bit of news this week.
[11:22] James.atLLOUD love me lots of chrome, but so many active processes
[11:23] George Equus: I went for Firefox, never give any trouble
[11:23] George Equus: Google is fine for searches..
[11:23] George Equus: IMHO
[11:23] Andrew.Hellershanks A new scripting engine feature has hit the httptests branch of OS. It is considered experimental at this point and is not reccommended for production use.
[11:24] Kayaker Magic: I'm all ears... tell us more.
[11:24] Ubit Umarov: even testing is only for the brave
[11:24] Andrew.Hellershanks um... take out that word feature. It is a new script engine. It is called MMREngine, IRC.
[11:25] Leighton.Marjoram Bill seems to like it on Opensim Life
[11:25] Ubit Umarov: you mean XMRengine i think :p
[11:25] Andrew.Hellershanks umm. Yeah, that is the name. :)
[11:25] Ubit Umarov: well it has several issues..
[11:26] Andrew.Hellershanks Not unlike the current scripting engine. :)
[11:26] Ubit Umarov: was great at avination but there it uses the version that requires a specially modified mono
[11:26] Ubit Umarov: without it its bad on windows.. and just seems to work on linux
[11:27] Andrew.Hellershanks When I was runing OS on Windows I didn't use mono to run instances.
[11:27] Ubit Umarov: mricker the author ( with melanie) made a few changes that im testing ( with my own)
[11:27] Ubit Umarov: does seem a little better now
[11:27] Ubit Umarov: but yeack.. lots of work..
[11:28] Ubit Umarov: and kayaker will not like it :p
[11:28] Kayaker Magic: Why not?
[11:28] Ubit Umarov: its major feature is handling long events a bit better
[11:28] Andrew.Hellershanks Also not advisable to mix it with regions using XEngine. You would need to recompile scripts on region crossings.
[11:29] Ubit Umarov: llSleep not a issue well or will not be
[11:29] Ubit Umarov: but they will lose cpu...
[11:29] Ubit Umarov: and will run again when possible
[11:29] Andrew.Hellershanks XMREngine is more SL like. It also has better memory management so no more need of the appdomains setting.
[11:29] Ubit Umarov: they can run more than the limit of 300ms of xengine...
[11:30] Ubit Umarov: but will run a bit slower
[11:30] Kayaker Magic: Sounds like they FIXED some problems with llSleep.
[11:30] Ubit Umarov: and of course all other events frozen by spec
[11:30] Andrew.Hellershanks Yes. It uses a better method of threading.
[11:31] Andrew.Hellershanks Ubit, I know of someone who pushed their script settings to allow almost 1 full second of script execution.
[11:31] Ubit Umarov: also during sleep the thread is released to do other work.. well or will be... it not on the version on git
[11:32] Ubit Umarov: another issue is that its parsing is more like SL...
[11:32] Ubit Umarov: that means come complex expressions will not give same result as Xengine
[11:32] Ubit Umarov: unless scripter was carefull using () etc
[11:33] Ubit Umarov: to enforce a specific parsing order
[11:33] Andrew.Hellershanks It also means scripts that use the old memsave hack will work.
[11:33] Ubit Umarov: hopefully should be a minor issue
[11:34] Ubit Umarov: for now not really working is a more relevant issue :P
[11:34] Ubit Umarov: it may also fix the memory use on script delete
[11:34] Andrew.Hellershanks Having the engine releasing a thread during sleep will be a big help.
[11:34] Ubit Umarov: it does seem to actually delete it
[11:34] Kayaker Magic: Parsing differences usually only have an effect when you write code that is too clever for its own good.
[11:35] Ubit Umarov: aurora script engine is very good also, does similar but does not release memory as Xengine doesn't
[11:36] Andrew.Hellershanks Kayaker, order of execution, no short-cutting in if's, and right to left execution can lead to issues of one isn't careful.
[11:36] Ubit Umarov: actually currently im trieng to make XMR more like aurora engine on execution control and filosofy
[11:36] Ubit Umarov: aurora/whitecore
[11:36] Andrew.Hellershanks nods
[11:37] Ubit Umarov: bc thats the only cross platform solution
[11:37] Ubit Umarov: we can not use modified mono versions
[11:37] Ubit Umarov: or even mono tasklets, not avaiable for windows
[11:37] Ubit Umarov: possible other platforms
[11:38] Ubit Umarov: anothe this is that it is LSL only
[11:38] Ubit Umarov: no c# java blabla xengine could do
[11:38] Ubit Umarov: this converts LSL directly to IL
[11:39] Ubit Umarov: well to AST then IL
[11:39] Ubit Umarov: details
[11:39] Andrew.Hellershanks Does anyone have any idea how many (if any) people use anything other than LSL with OS?
[11:39] Ubit Umarov: well no one can use really
[11:40] Ubit Umarov: not on a region open to the world
[11:40] James.atLLOUD you mean using ossl?
[11:40] Andrew.Hellershanks no. That would be very unsafe.
[11:40] Ubit Umarov: ossl is included on LSL i said
[11:40] Andrew.Hellershanks James, LSL and OSSL can be considered the same for this point.
[11:40] James.atLLOUD OK, sorry.
[11:40] Ubit Umarov: script language we use :)
[11:40] Andrew.Hellershanks np. Good to have the clarification out there.
[11:40] Kayaker Magic: But is anyone using c# instead of LSL?
[11:41] Ubit Umarov: a few on very special closed regions cases
[11:41] Ubit Umarov: hey can format the disk on a region script great :p
[11:42] Ubit Umarov: but bc a few do use is why i did not delete its option already
[11:42] Andrew.Hellershanks A C# based script could access system files.
[11:42] Kayaker Magic: Speaking of OSSL, I would like to use osKey2Name, it has a threat level of "low", but permissions "${XEngine|osslParcelO}ESTATE_MANAGER,ESTATE_OWNER" What exactly does that mean?
[11:43] Ubit Umarov: just never use it on normal regions
[11:43] Ubit Umarov: actually now in "$OSSL  :p
[11:43] Ubit Umarov: on httptests
[11:44] Andrew.Hellershanks Kayaker, that setting means the function can only be used by estae managers and estate owners, and possibly parcel owners if osslParcelO is set.
[11:44] Ubit Umarov: means the things defined on osslParcelO a few lines above plus ESTATE_MANAGER,ESTATE_OWNE
[11:45] Andrew.Hellershanks I'm trying to remember the difference between the lsl and ossl version of that function.
[11:45] Andrew.Hellershanks IIRC, the lsl one only works when the avatar is in the region but the os one always works
[11:46] Kayaker Magic: I see. There is an old SL way to get avatar name by doing a dataserver request which is a pain in the butt.
[11:46] Kayaker Magic: The OSSL function looks nice, but I cannot use it because of the permissions. This is typical of all the OSSL functions: I want to use them but dare not.
[11:47] Andrew.Hellershanks Kayaker, I think that required use of a special URL. You could do that in SL. I think I did provide a means of doing that via a URL for VH. I don't think it is generally available.
[11:47] Ubit Umarov: there was a fix on master.. hmm don't remember what lol
[11:48] Andrew.Hellershanks yea. Been ages since I've thought about that feature.
[11:48] Ubit Umarov: as you know.. at least for me httptests is the devel branch now
[11:48] Ubit Umarov: untill we do release a 0.9.0 postfixes
[11:48] Andrew.Hellershanks Ubit has been doing most of his current work in that branch.
[11:48] Ubit Umarov: and upgrade our jenkins
[11:49] Ubit Umarov: bc there is another thing about httptests
[11:49] Ubit Umarov: it does require .net 4.6
[11:49] Ubit Umarov: that means no more XP
[11:49] Ubit Umarov: vista still possible...
[11:49] Ubit Umarov: on linux that means mono 5.x
[11:50] Andrew.Hellershanks Lucky that I have that version installed.
[11:50] Ubit Umarov: possibly some late 4.x but ppl say those have issues
[11:50] Andrew.Hellershanks I will just need to remove a couple of env vars and/or symlinks so I will use the 5.x mono by default.
[11:50] Ubit Umarov: 4.6.2 i know does not support .net4.6
[11:51] Ubit Umarov: err or whatever the version was :p
[11:52] Ubit Umarov: im on and current mono version is 5.8 ??
[11:52] Andrew.Hellershanks ah, I have mono 5.4
[11:52] Ubit Umarov: well that may also be an issue for some
[11:52] Andrew.Hellershanks
[11:53] Ubit Umarov: but we can't stay on .net4.0
[11:53] Kayaker Magic: Ah, I just looked it up: To get the user name from UUID in SL, you called llRequestUsername and waited for a dataserver event. osKey2Name does the same thing in one call, but most people cannot use it.
[11:53] Andrew.Hellershanks The mono versions seem to be going up pretty rapidly over that last year
[11:53] Ubit Umarov: MS tring to fix it :p
[11:54] Andrew.Hellershanks Kayaker, you need to do a name look up from a UUID when the person isn't in the region?
[11:55] Kayaker Magic: Yes, if the person is in the region you don't need this. But to reliably get the name you need llRequestUsername.
[11:56] Ubit Umarov: btw viewers may also upload broken meshes even without analise and say nothing
[11:56] Ubit Umarov: but more rare
[11:57] Andrew Hellershanks: Kayaker, reliably? Hm... I can't say I've ever know of llKey2Name to fail.
[11:57] Ubit Umarov: and as i said 0.9 tries to refuse very trivial brokeness cases... a thing that make clever ppl say 0.9 is broken, as you know ;p
[11:58] Andrew Hellershanks: Ubit, better that than complaining that a mesh they uploaded isn't working.
[11:58] Ubit Umarov: yeah but tell that to some ppl.. whatever
[11:58] Andrew Hellershanks: yea, I know. :P
[11:58] Kayaker Magic: I had a case where I got the uuid of the owner of an object. The owner was not nearby, so lLKey2Name returned "". Ether I have to test for that and call llRequestUsername, or just give up and call llRequestUsername first and do it the hard way.
[11:59] Andrew Hellershanks: Right. If the user isn't nearby that is likely to fail.
[11:59] Kayaker Magic: Obviously if someone touches an object, I can assume they must be nearby and use llKey2Name. But there are other cases.
[11:59] Andrew Hellershanks: I have usually only needed to do the lookup in gadgets after someone touched an item.
[12:00] Andrew Hellershanks: Kayaker, I would be interested to know if llRequestUsername works. I have never tried using it.
[12:01] Kayaker Magic: It works fine, but you have to request and return. Wait for the dataserver event, then re-structure your code to pick up where it left off.
[12:01] Andrew Hellershanks: LSL_Api.cs code does issue two Async calls. One to get he information and the other to post it to the dataserver. Chances are it will work.
[12:01] George Equus: RL calling. Thanks for interesting evening.
[12:02] Andrew Hellershanks: Kayaker, that's pretty standard for a number of things where you use HTTP requests.
[12:03] Kayaker Magic: But for just getting a name? Pain in the butt.
[12:03] Andrew Hellershanks: Kayaker, chain your users down so they will always be in range. :)
[12:03] Kayaker Magic: OK, I'll try that!
[12:03] Andrew Hellershanks: hehe
[12:04] Kayaker Magic: But then they all have to have Restrained Love turned on....

Personal tools
About This Wiki