Chat log from the meeting on 2023-05-30

From OpenSimulator

Jump to: navigation, search
[11:03 AM PT]  Andrew Hellershanks: Hello, everyone.
[11:04 AM PT]  Andrew Hellershanks: I'm now on a new much higher speed Internet connection.
[11:04 AM PT]  Vincent.Sylvester Still four minutes late lol
[11:04 AM PT]  Gavin.Hird but some viewers have used the channel to distinguish the opensim build from the SL build, so for instance (no surpirse there) it is called FirestormOS-Release
[11:05 AM PT]  Vincent.Sylvester I could never find a standard for what strings go where with that
[11:05 AM PT]  Lyr Lobo: Hello *grins*
[11:05 AM PT]  Vincent.Sylvester Viewers send what they think is right or in some cases what the users set
[11:05 AM PT]  Gavin.Hird it is relatively messy
[11:06 AM PT]  Andrew Hellershanks: The company was here for an hour and a half messing about. Still not TV service. After that I needed to get something to eat.
[11:06 AM PT]  Ubit Umarov:
[11:07 AM PT]  Gavin.Hird so in any case just tweaking what is sent at login on the channel message to satisfy OSGs sensibilities
[11:07 AM PT]  Ubit Umarov: sorry gavin.. channel should inc a name, by spec :)
[11:07 AM PT]  Lyr Lobo: congrats on your speedy connection, Andrew
[11:08 AM PT]  Vincent.Sylvester I did talk to Dan a bit about the systems we have, messy is a good description. In checking I did notice a viewer reporting as libclient with a plaintext mac in the logs
[11:08 AM PT]  Gavin.Hird for a complex build setup like SL, yes, but it is totally uneccesary for a single viewer build setup
[11:08 AM PT]  Gavin.Hird makes things overly complicated
[11:09 AM PT]  Andrew Hellershanks: Lyr, thanks.
[11:09 AM PT]  Ubit Umarov: well we do
[11:09 AM PT]  Ubit Umarov: clientNameToCheck = channel   " "   clientVersion;
[11:10 AM PT]  Ubit Umarov: hmm except if version includes a spave.. if so we only use that... no idea legacy things i guess
[11:10 AM PT]  Web Rain: i got the llGetHTTPHeader grimlin, was my bad i forgot to allow the port 9000 in the localhost filter (private sandbox)... maybe the bug was that i recived something through the filter
[11:11 AM PT]  Vincent.Sylvester I was wondering why there was no mantis coming in about that, glad you got it solved
[11:12 AM PT]  Andrew Hellershanks: Web, happy to hear you got that issue solved. Now on to the next one.  ;)
[11:12 AM PT]  Web Rain: i did some experiments and all worked like a charm even with 50 requests with a timer set to 0.05
[11:12 AM PT]  Ubit Umarov: wel abotu new bugs... last week
[11:13 AM PT]  Ubit Umarov: Misterblue added one
[11:13 AM PT]  Ubit Umarov: well i mean code change not bug ;)
[11:13 AM PT]  Jamie.Jordan hi everybody
[11:13 AM PT]  Gavin.Hird Hi Jamie
[11:14 AM PT]  Ubit Umarov: we also had more changes related to httpclient use
[11:14 AM PT]  Andrew Hellershanks: Hello, Jamie
[11:14 AM PT]  Lyr Lobo: Welcome!
[11:15 AM PT]  Ubit Umarov: looking abotu that llgetheader issue i decided to make the search case insentive
[11:16 AM PT]  Ubit Umarov: by spec they should be lower case.. but.. with not look for them with any case ..
[11:16 AM PT]  Ubit Umarov: cpu cost on that is not that hig ( if ascII )
[11:16 AM PT]  Misterblue Waves: an UBit fixed it
[11:17 AM PT]  Web Rain: many headers return empty, i tested also the headers mentioned here
[11:17 AM PT]  Web Rain:
[11:18 AM PT]  Ubit Umarov: ahh yes the Extra constants for physics LSL did spam a lot on several regions per instance.. Not anymore
[11:18 AM PT]  Ubit Umarov: the clients do need to send them Web
[11:19 AM PT]  Ubit Umarov: and only send a few, depending o case...
[11:19 AM PT]  Vincent.Sylvester The explicit dispose for httpclient I still find quite odd given so many other libraries clean up after themselves without code telling them to
[11:19 AM PT]  Vincent.Sylvester Very manual compared to the old webrequest way
[11:19 AM PT]  Web Rain: i need just one a part the ip, is the "origin", pretty important for the verifications and safty
[11:20 AM PT]  Ubit Umarov: vicent old web did dispose the parts tthat needed it
[11:20 AM PT]  Ubit Umarov: and ity is abotu unmanaged or special resouces
[11:20 AM PT]  Ubit Umarov: webhttpclient had nothing do dispose
[11:20 AM PT]  Ubit Umarov: httpclient has..
[11:21 AM PT]  Ubit Umarov: ( on web.. we had to dispose the streams etc..
[11:21 AM PT]  Vincent.Sylvester Yeah essentially unwrapping that entire thing
[11:21 AM PT]  Ubit Umarov: now .. there is some automatic dispose
[11:22 AM PT]  Ubit Umarov: BUT
[11:22 AM PT]  Ubit Umarov: that it a lot slower
[11:22 AM PT]  Ubit Umarov: if you leave behind something with Dispose() and gc spots it is not in use... it flags it to look later
[11:23 AM PT]  Ubit Umarov: and addds extra processing on that..
[11:23 AM PT]  Ubit Umarov: so delayed and heavier
[11:23 AM PT]  Ubit Umarov: bc gc can't spot that is all free to release..
[11:23 AM PT]  Vincent.Sylvester Meaning in some respects httpclient might be faster or less heavy
[11:24 AM PT]  Ubit Umarov: wel httpclient is a viloation
[11:24 AM PT]  Ubit Umarov: violation
[11:25 AM PT]  Vincent.Sylvester I haven't gotten around to doing tests with it yet, no idea how it behaves, but I am cautious to say the least
[11:25 AM PT]  Vincent.Sylvester Supposedly was already in use before just wrapped
[11:25 AM PT]  Ubit Umarov: becasue the thing to dispose is actually a pool of network sockets
[11:25 AM PT]  Ubit Umarov: that should actually be shared
[11:26 AM PT]  Ubit Umarov: webhttprequesr had servicepoints
[11:26 AM PT]  Ubit Umarov: with aproper manager..
[11:26 AM PT]  Ubit Umarov: ms lost that.. doing a total mess
[11:26 AM PT]  Vincent.Sylvester As always
[11:27 AM PT]  Ubit Umarov: some of similar funtionality is there, on those pools per httpclient
[11:27 AM PT]  Ubit Umarov: so they say we should only allocate one for the entire app
[11:28 AM PT]  Ubit Umarov: but then that is a fail..  because each have parameters that are specific to some connectiosn and can not be shared
[11:28 AM PT]  Ubit Umarov: like Timeout |||
[11:28 AM PT]  Ubit Umarov: Timeout !!
[11:29 AM PT]  Ubit Umarov: their solution on that is to add a extra cancelation token qith a timeout
[11:29 AM PT]  Ubit Umarov: but that means 2 timers!!!
[11:29 AM PT]  Ubit Umarov: and only the shorter one will count..
[11:30 AM PT]  Vincent.Sylvester If I roll my eyes any harder I'll go blind heh
[11:30 AM PT]  Ubit Umarov: so i changed all the code again
[11:30 AM PT]  Andrew Hellershanks smiles at Vincent
[11:30 AM PT]  Ubit Umarov: now i do share the thing that as the pools
[11:31 AM PT]  Ubit Umarov: and make a new httpclient per connection
[11:31 AM PT]  Ubit Umarov: in fact what ms does on the IhttpclientFactory
[11:31 AM PT]  Ubit Umarov: fun how they recomend to do one thing and they do other...
[11:31 AM PT]  Ubit Umarov: whatever
[11:32 AM PT]  Vincent.Sylvester Where does this idea of pools == good come from? I see that in so many places and it almost always ends up a clunky mess
[11:32 AM PT]  Gavin.Hird resource constraints
[11:33 AM PT]  Joe Magarac: It's mostly an artifact of web servers which have a huge number of connections.
[11:34 AM PT]  Vincent.Sylvester The resource part I get, but implementation is often so poor it just stops working after a while
[11:34 AM PT]  Vincent.Sylvester Between http in the past eventually breaking down to dtl money and their db connection pools
[11:34 AM PT]  Gavin.Hird trix to sell more server capacity (Azure)
[11:35 AM PT]  Vincent.Sylvester Throwing hardware at software design issues, a classic
[11:35 AM PT]  Vincent.Sylvester Well in more positive news Tess has been reworking the artist and content creation sections on the wiki
[11:35 AM PT]  Orbert.Tatham If you are in the hardware business, you sell crap code.
[11:36 AM PT]  Orbert.Tatham Thank you tess
[11:36 AM PT]  Orbert.Tatham *Tess
[11:36 AM PT]  Ubit Umarov: pools are necessary to reuse sockets
[11:36 AM PT]  Gavin.Hird great stuff
[11:37 AM PT]  Ubit Umarov: without them, is is very easy to use all the allowed number of openfiles
[11:37 AM PT]  Ubit Umarov: because of tpc TIME_WAIT
[11:37 AM PT]  Ubit Umarov: and things like that
[11:37 AM PT]  Misterblue Waves: There was some noise that LL was going to some OAuth or 2FA stuff for SL. Does anyone know where that stands?
[11:37 AM PT]  Ubit Umarov: also needed to do http keepalive and http 1.1 pipeline
[11:38 AM PT]  Gavin.Hird done a year ago
[11:38 AM PT]  Gavin.Hird I have not heard anything new to happen in that respect lately
[11:38 AM PT]  Vincent.Sylvester 2FA lol they should allow longer passwords that might actually help
[11:39 AM PT]  Joe Magarac: Question about how connections start.
[11:39 AM PT]  Andrew Hellershanks: Go ahead, Joe.
[11:39 AM PT]  Joe Magarac: At startup, login reply tells viewer what regin.
[11:40 AM PT]  Joe Magarac: region.
[11:40 AM PT]  Joe Magarac: But not where in the region.
[11:40 AM PT]  Joe Magarac: Viewer sends an AgentReply over UDP.
[11:40 AM PT]  Joe Magarac: That contains CameraCenter.
[11:40 AM PT]  Joe Magarac: Until that's sent, server does not send object updates.
[11:40 AM PT]  Joe Magarac: After AgentReply has been sent, server starts sending agent updates.
[11:41 AM PT]  Joe Magarac: sorry, Object Updates.
[11:41 AM PT]  Joe Magarac: The first ObjectUpdate for the avatar tells the viewer where the avatar is.
[11:42 AM PT]  Joe Magarac: Then its next AgentUpdate tells the server where the camera is.
[11:42 AM PT]  Joe Magarac: This is a sudden jump from whatever dummy camera value was used in in the initial agent update.
[11:43 AM PT]  Joe Magarac: This jump messes up SL servers and seems to be the cause of some interest list bugs.
[11:43 AM PT]  Gavin.Hird there is no point ins storing the region location on the viewer side on login, because the avatar migh have been loged in elsewhere and are actually in a completely different place
[11:43 AM PT]  Gavin.Hird so ti has to come from the server
[11:43 AM PT]  Joe Magarac: Correct.
[11:43 AM PT]  Joe Magarac: The problem is that chicken and egg problem.
[11:43 AM PT]  Joe Magarac: Server can't send object updates until it has a camera location.
[11:44 AM PT]  Ubit Umarov: don't think there is such relation
[11:44 AM PT]  Joe Magarac: Viewer doesn't know camera location until it gets an object update for the avatar.
[11:44 AM PT]  Ubit Umarov: but my memory is well known to be very bad :)
[11:44 AM PT]  Gavin.Hird why would the camera location be anywhere else than the agent at login?
[11:45 AM PT]  Joe Magarac: Landing point. Ban line. Landmark.
[11:45 AM PT]  Gavin.Hird I don't think I have ever logged in with the cam being elsewhere?
[11:45 AM PT]  Joe Magarac: The viewer doesn't know where those are until the server tells it.
[11:45 AM PT]  Ubit Umarov: avatar is told to be created on a position actually
[11:45 AM PT]  Orbert.Tatham The viewer doesn't know where that is until itis told
[11:45 AM PT]  Gavin.Hird don't see a problem with that
[11:45 AM PT]  Ubit Umarov: either by  grid, or by sender region on tp
[11:46 AM PT]  Ubit Umarov: and it can change it, for example doing landpoint
[11:46 AM PT]  Joe Magarac: Server won't send object updates until it receives the first AgentUpdate.  At least, SL servers won't.
[11:46 AM PT]  Joe Magarac: I tried.
[11:46 AM PT]  Gavin.Hird all that can have cahnged from the last viewer login instance on a machine if the avatar is logged in on another viewer instance, so all that must come from the server
[11:46 AM PT]  Joe Magarac: Right.
[11:47 AM PT]  Joe Magarac: The problem is this.
[11:47 AM PT]  Joe Magarac: Viewer sends agent update. Server starts sending object updates based on that location.
[11:47 AM PT]  Joe Magarac: Viewer gets object update for avatar.
[11:47 AM PT]  Joe Magarac: Viewre sends agent update with new  location. This changes what's "interesting".
[11:48 AM PT]  Joe Magarac: Now the server has to abort sending the old items of interest and recalculate from the new camera position.
[11:48 AM PT]  Joe Magarac: This seems to occasionally resuilt in a sim/viewer out of sync condition.
[11:48 AM PT]  Joe Magarac: And some objects don't appear.
[11:49 AM PT]  Joe Magarac: At least for SL servers.
[11:49 AM PT]  Joe Magarac: How does wori in Open Simulator?
[11:49 AM PT]  Gavin.Hird because of the changed interest list you mean?
[11:49 AM PT]  Ubit Umarov: wel avatar initial position is decided by server
[11:49 AM PT]  Joe Magarac: Right.
[11:49 AM PT]  Ubit Umarov: ( landpoints.. telehubs. bans.. etc )
[11:49 AM PT]  Joe Magarac: Right.
[11:50 AM PT]  Gavin.Hird why do the server even care about where the avatar thinks it is on login. The server is the authorative source of where the avatar is
[11:50 AM PT]  Ubit Umarov: yeah,..  i just said that :)
[11:50 AM PT]  Joe Magarac: Because the list of objects to be sent depends on camera location and direction and far plane and viewing frustrum.
[11:50 AM PT]  Gavin.Hird unless you specify a new login region that is different from logout location
[11:50 AM PT]  Ubit Umarov: avatar position is Server side
[11:50 AM PT]  Ubit Umarov: camera position is viewer side
[11:50 AM PT]  Gavin.Hird or you log on via a LM
[11:51 AM PT]  Joe Magarac: Yes.
[11:51 AM PT]  Gavin.Hird or slurl too IÂ suppose
[11:51 AM PT]  Ubit Umarov: well don't tell..  but bc issues that that, opensim only does priority by avatar position.. not camera :p
[11:51 AM PT]  Orbert.Tatham Even under those circmstances, the viewer is only guessing at where htte camera is until it is told
[11:52 AM PT]  Gavin.Hird but is this a real perceived issue at all?
[11:52 AM PT]  Misterblue Waves: the server sends different objects to the viewer depending on where the camera is
[11:52 AM PT]  Joe Magarac: Oh, yes. It causes random objects to not be sent, at least for SL servers.
[11:52 AM PT]  Ubit Umarov: priority by camera is a big mess .. does tons of reprio..  by jsut rotating the avatar around
[11:52 AM PT]  Misterblue Waves: so If the use zapps to a different location, the sent objects list gets regenerated/reordered
[11:53 AM PT]  Ubit Umarov: well.. currently we do send ALL objects
[11:53 AM PT]  Joe Magarac: Oh.
[11:53 AM PT]  Ubit Umarov: jusr retune the order as avatar moves
[11:53 AM PT]  Gavin.Hird I think the random objects not being sent on SL server has more to do with latency and lag in the CDN
[11:53 AM PT]  Ubit Umarov: SL moved culling etc to viewer side
[11:53 AM PT]  Ubit Umarov: when they added viewer side cache
[11:53 AM PT]  Joe Magarac: No, it's not the CDN. In my own viewer I can tell.
[11:53 AM PT]  Ubit Umarov: sl viewers do want ALL prims
[11:54 AM PT]  Ubit Umarov: no server side culling
[11:54 AM PT]  Joe Magarac: Only those within draw distance.
[11:54 AM PT]  Ubit Umarov: no
[11:54 AM PT]  Ubit Umarov: it wants all
[11:54 AM PT]  Gavin.Hird In any case , the AIS update is soon ready, so all of this behaviro will most likely change in SL
[11:54 AM PT]  Ubit Umarov: do do objects cache
[11:54 AM PT]  Misterblue Waves: OS sends all objects but sorts the order
[11:54 AM PT]  Joe Magarac: That makes sense.
[11:54 AM PT]  Gavin.Hird the are only waiting for some server side changes to be ready to go
[11:54 AM PT]  Joe Magarac: Even past the "far" distance?
[11:55 AM PT]  Ubit Umarov: i added server  side culling..  but disabled, bc of that
[11:55 AM PT]  Vincent.Sylvester If something doesn't load usually most you have to do is turn around once and they pop back in
[11:55 AM PT]  Gavin.Hird yes Vincent. My experience too
[11:55 AM PT]  Joe Magarac: When this happens in SL, you have to leave the area and come back.
[11:56 AM PT]  Ubit Umarov: you are doing viewer objects cache like sl ?
[11:56 AM PT]  Gavin.Hird could also be that some people use overly large caches, which slows down the process
[11:56 AM PT]  Vincent.Sylvester Version of FS I have has object culling set pretty aggressively so I see objects pop in mostly 50/50 cache and server
[11:56 AM PT]  Ubit Umarov: like asking for probes...
[11:56 AM PT]  Ubit Umarov: then ask for updates on changed things..
[11:57 AM PT]  Joe Magarac: Actually, I don't have a viewer object cache yet. I ask for the full interest list.
[11:57 AM PT]  Ubit Umarov: ok
[11:57 AM PT]  Joe Magarac: I have mesh and texture caches.
[11:57 AM PT]  Joe Magarac: But not object update caches.
[11:57 AM PT]  Web Rain: It reminds me a bit the the lookAt of osTeleportAgent... so a tp at login maybe a solution but have to provide the lookAt stored on the logout in the viewer side
[11:57 AM PT]  Ubit Umarov: so here as  we said you will get all objects on  a ordered ( and reorder) way
[11:58 AM PT]  Ubit Umarov: based on aparent size rule
[11:58 AM PT]  Joe Magarac: Yes. That makes sense.
[11:58 AM PT]  Ubit Umarov: well very aproximate apparent size
[11:58 AM PT]  Gavin.Hird whose cache code are you using to implement your viewer?
[11:58 AM PT]  Joe Magarac: Mine.
[11:58 AM PT]  Joe Magarac: All new code, 100% safe Rust.
[11:59 AM PT]  Gavin.Hird ok, but you must look at some spec / code
[11:59 AM PT]  Ubit Umarov: there are flags you need to send to tell region to not send you sl cache things
[11:59 AM PT]  Joe Magarac: I look mostly at Firestorm code.
[11:59 AM PT]  Gavin.Hird the FS cache code is not the same as the SL viewer code
[11:59 AM PT]  Joe Magarac: Yes, I send the flag that asks for all objects un-cached.
[11:59 AM PT]  Joe Magarac: Right. So where is this documented?
[11:59 AM PT]  Gavin.Hird FS donnated code to LL who modded it a lot
[12:00 PM PT]  Ubit Umarov: if(viewerCache 
[12:00 PM PT]  Gavin.Hird not sure of FS backported all those mods
[12:00 PM PT]  Ubit Umarov: this is original updates send
[12:00 PM PT]  Gavin.Hird So I would look at the official SL code fist and foremost
[12:00 PM PT]  Gavin.Hird first*
[12:00 PM PT]  Ubit Umarov: with wrong flags you will get porobes not full updates
[12:00 PM PT]  Joe Magarac: OK, well that would send everything.
[12:01 PM PT]  Ubit Umarov: opos thats about prims :)
[12:01 PM PT]  Ubit Umarov: bool cacheCulling = (flags 
[12:02 PM PT]  Ubit Umarov: this are the viewer flags..
[12:02 PM PT]  Joe Magarac: Right. Checking my code...
[12:02 PM PT]  Gavin.Hird I hardly ever see the behavior you describe but there are discussion in the SL viewer forum all the time that FS users may have such issues.
[12:03 PM PT]  Joe Magarac: Right. OK, I'm satisfied this will work OK in OS.
[12:03 PM PT]  Ubit Umarov: those you send on regionhandshake packet
[12:03 PM PT]  Vincent.Sylvester Spend enough time in here and the glitches become second nature, you get worried when they suddenly disappear :)
[12:04 PM PT]  Ubit Umarov: RegionInfo.Flags  on it
[12:04 PM PT]  Gavin.Hird :-)
[12:04 PM PT]  Joe Magarac: Right.
[12:04 PM PT]  Joe Magarac: Bye for now. SL meeting starting.
[12:04 PM PT]  Gavin.Hird Bye
[12:04 PM PT]  Andrew Hellershanks: ok, Joe. See you again another day.
[12:04 PM PT]  Ubit Umarov: btw that is a mandatory packet
[12:05 PM PT]  Gavin.Hird we don't have any glitches - what are you insinuating? Only undocumented features.
[12:05 PM PT]  Tess JL: Of course!
[12:05 PM PT]  Andrew Hellershanks: We are just past the top of the hour. Is there anyone else with questions or comments for todays gathering?
[12:05 PM PT]  Vincent.Sylvester Speaking of
[12:05 PM PT]  Tess JL: Just a quick one I hope
[12:05 PM PT]  Andrew Hellershanks: Tess, ok. You will be after Vincent with his "speaking of"
[12:05 PM PT]  Tess JL: okie
[12:06 PM PT]  Vincent.Sylvester I was going through lsl code and saw that we have specific code for playing sounds on links
[12:06 PM PT]  Vincent.Sylvester Different from the other code that handles links it plays on root part or specific links only
[12:06 PM PT]  Vincent.Sylvester LINK_SET is not pushing to all links
[12:06 PM PT]  Vincent.Sylvester I couldn't find whether it is meant to or not
[12:06 PM PT]  Vincent.Sylvester lsl wiki just states that the constant is supported, but not how it behaves
[12:07 PM PT]  Vincent.Sylvester If someone could test, I don't have SL
[12:07 PM PT]  Vincent.Sylvester That's all :)
[12:08 PM PT]  Andrew Hellershanks: In the case of llMessageLinked, LINK_SET sends to all prims
[12:08 PM PT]  Ubit Umarov: playsounf on all prims is.. suicidal
[12:08 PM PT]  Tess JL: If it doesn't, it's a bug
[12:09 PM PT]  Tess JL: llLinkPlaySound() is a weird one anyway. Seems LL wanted it to replace jsut about every old sound player function
[12:09 PM PT]  Ubit Umarov: yeah i only do it dor 1 prim
[12:10 PM PT]  Andrew Hellershanks: Any further comments on llPlaySound?
[12:10 PM PT]  Ubit Umarov: fun to linkset on 1000 prims set :p
[12:11 PM PT]  Andrew Hellershanks: hehe, yea. That wouldn't be very good.
[12:11 PM PT]  Andrew Hellershanks: Ok, Tess. What do you have for us today?
[12:12 PM PT]  Tess JL: It's about the sound script functions actually ;-)
[12:12 PM PT]  Andrew Hellershanks: :)
[12:13 PM PT]  Tess JL: I've been doing some tweaks to the content creation part of the wiki and come across something I'm not sure how to handle. What do we do about lsl functions taht duplicate osssl functions exactly?
[12:13 PM PT]  Ubit Umarov: why i hell does ll allow link_set on that?
[12:13 PM PT]  Ubit Umarov: jezz
[12:13 PM PT]  Tess JL: I suppose we should recommend one over the other but which?
[12:13 PM PT]  Ubit Umarov: don't think we have many of those
[12:13 PM PT]  Ubit Umarov: specially the "exact" part
[12:14 PM PT]  Ubit Umarov: we did had setparameters*
[12:14 PM PT]  Tess JL: There are three sound funtions I found
[12:14 PM PT]  Ubit Umarov: well sl did not had the link ones
[12:14 PM PT]  Tess JL: llCollisionSound() / osCollisionSound()
[12:14 PM PT]  Tess JL: llLinkAdjustSoundVolume() / osAdjustSoundVolume()
[12:14 PM PT]  Ubit Umarov: those are a bif dof
[12:14 PM PT]  Vincent.Sylvester Usually the ossl just adds some that is missing from lsl, parameters or functions that have gets and no sets that sort of thing
[12:14 PM PT]  Ubit Umarov: i think
[12:14 PM PT]  Tess JL: llLinkSetSoundRadius() / osSetSoundRadius()
[12:14 PM PT]  Andrew Hellershanks: Tess, if they really are the same I would favour the ll functions as they would be portable between here and other grids.
[12:15 PM PT]  Vincent.Sylvester Use ossl only when needed
[12:15 PM PT]  Tess JL: Ok
[12:15 PM PT]  Vincent.Sylvester Avoid if possible, ends up being more compatible
[12:15 PM PT]  Tess JL: It seems that what happened is the opensim added a function first, the LL got the same idea later
[12:15 PM PT]  Vincent.Sylvester and implemented it poorly
[12:15 PM PT]  Web Rain: there is always a difference in terms of limitations... ll and os are not the same...
[12:15 PM PT]  Andrew Hellershanks: Tess, that's possible.
[12:16 PM PT]  Web Rain: blessed be for the strings manipulations functions ^^
[12:16 PM PT]  Andrew Hellershanks: Web, that is often one of the differences between the ll and os versions.
[12:16 PM PT]  Tess JL: There's also that weird llLinkPlaySound VIncent mentioned. It's almost but not quite the same as osPlaySound()
[12:16 PM PT]  Tess JL: Anyway, ll before os. Got it!
[12:17 PM PT]  Vincent.Sylvester Some ossl also exists to make something less complex to use, a single function rather than having two nest them, especially for links
[12:18 PM PT]  Andrew Hellershanks: Tess, anything else?
[12:18 PM PT]  Ubit Umarov: some was made before LL had similar funtions
[12:18 PM PT]  Ubit Umarov: and may still be diferent
[12:18 PM PT]  Vincent.Sylvester Them stealin our stuff ain't nothin new
[12:18 PM PT]  Ubit Umarov: like the sound ones only work on one prim at a time
[12:19 PM PT]  Ubit Umarov: ( well but a made the ll one also only do that ;) )
[12:19 PM PT]  Tess JL: A lot Andrew but I think I better keep quiet for now or we won't be finished until next week
[12:19 PM PT]  Vincent.Sylvester Tess if you have questions about how those things work in code I'll happily go through it and let you know
[12:19 PM PT]  Vincent.Sylvester Gives me a reason to look for bugs or optimizations in there
[12:20 PM PT]  Andrew Hellershanks: Tess, ok. :)  Next week remind me that you have some topics so we can get some of your others issues addressed early.
[12:20 PM PT]  Tess JL: I don't think that's necessary. I'm trying to write simple, practical explanations and leave the under-the-hood stuff ot the experts
[12:20 PM PT]  Tess JL: Okie, Andrew
[12:21 PM PT]  Misterblue Waves: speaking of LSL functions, what about
[12:21 PM PT]  Andrew Hellershanks: Tess, while you are working on the wiki pages, keep an eye out for pages with images where those images are hosted somewhere other than on the machine.
[12:21 PM PT]  Misterblue Waves: those are SL functions that aren't in OS
[12:21 PM PT]  Vincent.Sylvester Easy, no
[12:22 PM PT]  Tess JL: Ok, I will
[12:22 PM PT]  Vincent.Sylvester LL smoking the dryer lint when they came up with that linkset data crap
[12:22 PM PT]  Vincent.Sylvester We talking about those functions on past meetings already at length
[12:23 PM PT]  Tess JL: Ummm... I think some earlier opensim developer thought that "linkset crap" was a good idea too
[12:23 PM PT]  Misterblue Waves: I agree that it's LL cruft, but if we're more or less SL compatible, we should have their cruft also
[12:23 PM PT]  Vincent.Sylvester You gonna get Ubit's blood pressure into the dangerous levels
[12:23 PM PT]  Lyr Lobo grins
[12:24 PM PT]  Andrew Hellershanks: oh, that linkset data stuff. I'm not sure we really want that.
[12:24 PM PT]  Vincent.Sylvester There are numerous things wrong with it, I only touched a few in my comments on mantis, there is more to that iceberg
[12:24 PM PT]  Vincent.Sylvester Regex...
[12:24 PM PT]  Vincent.Sylvester Even the accompanying functions for the string replace stuff they added for that stuff are a mess
[12:25 PM PT]  Vincent.Sylvester A pattern replace with just a count value, no start or end index, silly
[12:27 PM PT]  Andrew Hellershanks: Almost half past the hour now. Any last minute items? I know Ubit may have to be going afk soon.
[12:28 PM PT]  Web Rain: i use osRegexIsMatch and osReplaceString and works fine
[12:28 PM PT]  Web Rain: yes just an update
[12:28 PM PT]  Andrew Hellershanks: Go ahead, Web
[12:29 PM PT]  Tess JL: Just a quick one, Ubit, I'll IM you about any changes I do to the script function pages on the wiki so you can check I haven't done anything wrong
[12:29 PM PT]  Web Rain: I have a quick question about scripting security. is there an evolution on this subject and can we put sensitive data such as API keys in a script?
from what I know we can do it in a standalone / HG with precautions but not in the grids, is this still the case?
[12:29 PM PT]  Selby.Evans bye all
[12:29 PM PT]  Andrew Hellershanks: ok, Bye, Selby. Have a good week.
[12:29 PM PT]  Vincent.Sylvester Depends on whether the grid is open or closed and who ultimately owns the regions within
[12:30 PM PT]  Andrew Hellershanks: Depends on how you send the use and/or send the sensitive key data.
[12:30 PM PT]  Vincent.Sylvester API management in OpenSim you need to be proactive about and check who is using the keys and set limits etc.
[12:31 PM PT]  Web Rain: so, is still the same... thank you for the infos and works ^^
[12:31 PM PT]  Vincent.Sylvester Incoming requests should contain some data like IP at the very least, using request url you can also directly run API to send to regions rather them requesting data
[12:32 PM PT]  Ubit Umarov: ok i need go rl
[12:32 PM PT]  Ubit Umarov: cya later
[12:32 PM PT]  Tess JL: Me too
[12:32 PM PT]  Tess JL: Bye for now
[12:32 PM PT]  Gavin.Hird later
[12:32 PM PT]  Andrew Hellershanks: ok, I'll call todays meeting to a close.
[12:32 PM PT]  Andrew Hellershanks: Thank you all for coming. See you again next week.
Personal tools
About This Wiki