Simulator User Group/Transcripts/2011.04.08
|Prev 2011.04.05||Next 2011.04.12|
List of Speakers
|Andrew Linden||Arawn Spitteler||Ardy Lay|
|DogWomble Dollinger||Draconis Neurocam||Fancy Greeter|
|Flip Idlemind||JeanMark Nightfire||Kadah Coba|
|Kelly Linden||Liisa Runo||Maestro Linden|
|Moundsa Mayo||Mudslinger Ning||Pauline Darkfury|
|Sigma Avro||Simon Linden||Stickman|
|Tankmaster Finesmith||Techwolf Lupindo|
[15:57] DogWomble Dollinger: hi simon
[15:57] JeanMark Nightfire: greetings
[15:58] Simon Linden: Hello everyone
[15:58] Draconis Neurocam: hello simon
[15:58] Moundsa Mayo: Allo, all!
[15:58] DogWomble Dollinger: change of outfit simon, or am i just imginign things?
[15:59] Simon Linden: It's been a few weeks, but I did finally change my AV a bit
[15:59] Moundsa Mayo: Oops - 'scuse - did i take your seat Simon?
[15:59] Moundsa Mayo: (I didn't check the numbers B^)
[15:59] Simon Linden: There's always more :) I don't think we've ever maxxed out the table
[16:00] Moundsa Mayo: Ooohh, sounds like a fun game!
[16:00] Simon Linden: I wonder what 12 scripts I have on
[16:00] Tankmaster Finesmith: looks good, simon
[16:00] Moundsa Mayo: Oh, yeah, my Mystitool. Need a slimmer gadget - only use three features.
[16:00] Draconis Neurocam: resizers
[16:00] Simon Linden: ah, some resizers
[16:03] Kadah Coba attaches more scripts to raise his score
[16:03] Simon Linden: hehe ... you'll win until you try a TP or region crossing :)
[16:04] Kadah Coba: I've region crossed with 6000k mono scripts recently
[16:04] DogWomble Dollinger: kadah, you were able to move about 3 days later?
[16:04] DogWomble Dollinger: :P
[16:04] Moundsa Mayo: Surprisingly, i usually make it fine over region crosings, even in fast vehicles. Never had as frequent a problem as others report.
[16:04] Moundsa Mayo: But I'm pretty much a vanilla avatar, otherwise.
[16:04] Kadah Coba: One of the RC (I think it was mag) and release channel has/had really nice region crossing recently :>
[16:05] Kadah Coba: The others, not so much
[16:05] Draconis Neurocam: is andrew in or will it just be you simon?
[16:05] Moundsa Mayo: A little overweight, but I don't think that's an obstacle at region boundaries.
[16:05] Simon Linden: Hmm, I'll ping Andrew
[16:05] Moundsa Mayo: 'Just' Simon? He's a gracious plenty!
[16:05] Simon Linden: ah, he's out today with the flu
[16:05] Draconis Neurocam: indeed
[16:05] Draconis Neurocam: ah shame
[16:05] Simon Linden: So send him lots of IMs :)
[16:06] Draconis Neurocam: well what have you been up to?
[16:06] Tankmaster Finesmith: oskar is onine, maybe he can grace us with his presence ?
[16:06] Simon Linden: Well, I'll try to run it somewhat organized as he does...
[16:06] Moundsa Mayo: Izzat a new flu cure? Should tell NIH and CDC!
[16:06] Kadah Coba: Dont send IM, send tp lures, hundreds of them. They'll be bound to accidentally accept one :P
[16:06] Tankmaster Finesmith: lol
[16:06] Simon Linden: For annoucements, the server rolls this week were smooth
[16:06] Pauline Darkfury: lol
[16:07] Tankmaster Finesmith: is there a limit to how hemy tp lures the server will store when the user is offline?
[16:07] Tankmaster Finesmith: meny*
[16:08] Simon Linden: I'm not sure what was moved into the RC channels this week ... I think "Magnum" got a new project
[16:08] Simon Linden: That is Monty Linden's "Fast Assets" project
[16:08] Arawn Spitteler: What's a TP Lure?
[16:08] Flip Idlemind: When you TP someone
[16:08] Simon Linden: He's been looking at the data pipeline between our simulators and the back-end asset system
[16:08] JeanMark Nightfire: sending someone a teleport offer
[16:08] Arawn Spitteler: Monty Linden scripts in Python?
[16:08] Moundsa Mayo: An unannounced TP invitation promising many L$
[16:09] Simon Linden: Delays there show up in your asset loading time ... slower textures, object rezzing, etc
[16:09] Moundsa Mayo: LOL, Arawn B^D
[16:09] Arawn Spitteler: I wasn't aware that any such Invites could be preserved for someone off line
[16:09] Simon Linden: I don't know much about the details, but I think he's put in some more metrics to measure performance better and found a few things to speed it up
[16:09] JeanMark Nightfire: speeding up is good
[16:10] DogWomble Dollinger smiles at "speed it up"
[16:10] Tankmaster Finesmith: yes, speed is always good...
[16:10] Moundsa Mayo: Yah, and really looking forward to llTeleportVehicleWithPassengers()!
[16:10] Mudslinger Ning: oooh- vechile teleports
[16:10] Simon Linden: Andrew has been split between the mesh project and his usual bug fixing
[16:10] Tankmaster Finesmith: :o
[16:11] Simon Linden: haha, that would be a good one
[16:11] Simon Linden: We do need to overhaul how vehicle region crossings work ... they are pretty ugly under the hood
[16:11] Sigma Avro: or llTeleportVehicleWithoutAgent()
[16:11] Simon Linden: Not to mention ugly in-world
[16:11] Pauline Darkfury: Vehicle TP would be adored by certain sci-fi vehicle builders (notably TARDIS & SG stuff, but also others)
[16:11] JeanMark Nightfire: yes, i have tried to install trolleys in our sims... doesnt work...
[16:11] Ardy Lay: llBackToTheFuture();
[16:12] Sigma Avro: lol
[16:12] Draconis Neurocam: 88 miles per hour
[16:12] Ardy Lay: I want to build a time machine!
[16:12] Pauline Darkfury: We've already got llSendVehicleToRandomPlaceWithoutAgents() ;)
[16:12] Ardy Lay: Hehe
[16:12] Simon Linden: I've been putting the final touches on llRegonSayTo() today and hope to get it into testing early next week
[16:12] Mudslinger Ning: heh - it might just be perfect for my cheetah flux backpack
[16:12] Moundsa Mayo: Yah, TeleportVehicleWithPassengers will add a lot to intergalacti travel!
[16:12] Ardy Lay: Thanks Simon, I need that one.
[16:12] Arawn Spitteler: llCrossBoundarySansAgents() works pretty good, as far as I can see
[16:12] Draconis Neurocam: nice, regionsayto is going to be great
[16:12] Kadah Coba: I'm sure the self replicating cubes would love the ability to TP around :3
[16:13] Flip Idlemind: Can "overhaul how vehicle region crossings work" be put on someone's official to do list?
[16:13] Pauline Darkfury: It could be restricted to vehicles with attached AVs
[16:13] Moundsa Mayo: Yah, Arawn, that one's been functional for a while, so we're partway there already B^)
[16:13] Simon Linden: We definitely know about the problems, Flip. I'm not sure when we'll get to it, however
[16:13] Mudslinger Ning: yeah - limited to TP with an avvie attached should be a good limiter on greifer effects
[16:14] Moundsa Mayo: llTVWP could be slow intentionally, show a 'hyperspace' animation while the servers do their thing.
[16:14] Pauline Darkfury: with no AVs, it can be simulated by having a vehicle rezzer at the far side of the portal/wormhole anyway
[16:14] Pauline Darkfury: Yeah, swirly glowing vortices for 30s would not upset people, if that was needed
[16:14] Fancy Greeter: Maestro Linden has arrived! (Or, returned?)
[16:15] DogWomble Dollinger: hi maestro
[16:15] Draconis Neurocam: hello maestro
[16:15] Flip Idlemind: There is a limit to how frequently an avatar can TP. If you are going to allow objects to teleport you could probably just use the same limit
[16:15] Moundsa Mayo: But we do appreciate all the recent enhancements, for sure!
[16:15] Sigma Avro: have the feeling that back and forth from different channels is not symetrical in time
[16:15] Maestro Linden: hi guys
[16:15] Tankmaster Finesmith: welcome maestro
[16:15] Fancy Greeter: Kelly Linden has arrived! (Or, returned?)
[16:15] Simon Linden: The final changes I made today to llRegionSayTo() will make it work better with AVs and attachments
[16:15] Moundsa Mayo: Allo Maestro!
[16:15] Mudslinger Ning: I think I'll make my own "whoosh" effects for my vechiles
[16:15] DogWomble Dollinger: hi kelly
[16:16] Simon Linden: Kelly or Maestro - do you have any annoucements or topics before we open it up?
[16:16] Pauline Darkfury: nice to hear llRegionSayTo is coming along nicely. I take it that individual attachments, as well as just the AV key will be targettable?
[16:16] Maestro Linden: we could talk about recent changes in Mono2Performance
[16:16] Moundsa Mayo: Hey Kelly!
[16:16] Pauline Darkfury: Hi Kelly, Maestro :)
[16:16] Simon Linden: Pauline - yes, here are the rules:
[16:16] Kelly Linden: I suppose we could Maestro.
[16:16] Ardy Lay: Instead of a black screen when logging in or teleporting we need someything interesting to stare at, like ElectricSheep or something.
[16:16] DogWomble Dollinger: maestro, yesplz :) that's something that interests me
[16:16] Simon Linden: llRegionSayTo() won't work on DEBUG_CHANNEL, those are just dropped
[16:17] DogWomble Dollinger: electric sheep? i want hippos!
[16:17] Draconis Neurocam: will it work on channel 0 simon?
[16:17] Simon Linden: Messages sent directly to an object (normal or attachment) just go to that prim
[16:17] Simon Linden: They can be sent on any channel, but aren't heard by other objects
[16:17] Draconis Neurocam: interesting
[16:18] Maestro Linden: afk 2 minutes - need to ask somebody downstairs something
[16:18] Simon Linden: If sent to an AV on channel 0, it is sent down to the viewer and will appear in a new color (assuming you get the viewer with those changes)
[16:18] Tankmaster Finesmith: k
[16:18] Moundsa Mayo: To list?
[16:18] Simon Linden: If sent to an AV not on channel 0, it can be heard by any attachment on the AV listening on the proper channel
[16:18] Moundsa Mayo: Or single ID?
[16:18] Flip Idlemind: Will you not get the message if you don't have the right viewer, or will it just not be special-colored
[16:18] Simon Linden: It's to a single ID
[16:18] Moundsa Mayo: k
[16:19] Moundsa Mayo: Delay?
[16:19] Moundsa Mayo: Or is ther a wiki page already - looking ...
[16:19] Simon Linden: I tested our 2.x viewer without my changes, and 1.23, and you get the message but the color is the default
[16:19] Simon Linden: The costs are the same as llSay()
[16:19] Simon Linden: (and I forget what those are)
[16:19] Draconis Neurocam: none really
[16:20] Liisa Runo: no delay
[16:20] Pauline Darkfury: Pretty sure llSay and llWhisper are relatively unrestricted
[16:20] Flip Idlemind: Is it limited to the region (probably, since it has "Region" in the name) or will it work on "child agents"
[16:20] Moundsa Mayo: Cool.
[16:20] Simon Linden: Yes, I built a griefer object and sent 100 messages to the toucher ... it's just like llSay()
[16:20] Pauline Darkfury: can certainly be spammy and send 600 or so llWhispers quickly (e.g. something like hippoINVENTORY listing a lot of contents)
[16:20] Draconis Neurocam: this would work as a nice in sim workaround for instantmessage then
[16:21] Simon Linden: It's limited to the region ... going beyond that was going to make it much more complex, so I went with the simple and faster approach
[16:21] Pauline Darkfury: Yeah, seems like it's a good alternative to llIM when you don't care about off-sim
[16:21] Sigma Avro: Intended to be implemented soon ?
[16:21] Moundsa Mayo: Is this perhaps a precursor to llOtherRegionSayTo?
[16:21] Pauline Darkfury: Will it work to people in the small distance outside the sim where they are visible to llSensor and the like (34m?)
[16:21] Simon Linden: Exactly ... it can obviously be used for object-to-object communication, but could also be useful for a greeter object, for example
[16:22] Kelly Linden: Why not a whole lot of functions with specific region names so you can have llDenbySayTo and llMorrisSayTo ....
[16:22] Simon Linden: A greeter could send it's message without spamming others in the area
[16:22] Kadah Coba: lol, Kelly
[16:22] Draconis Neurocam: have a regioncorner variable
[16:22] Draconis Neurocam: haha
[16:22] Simon Linden: llParcelSayTo() ?
[16:23] Sigma Avro: or set height limits for llSay ?
[16:23] Pauline Darkfury: Mmm, might be some usefulness for llParcelSay and llParcelSayTo, if that would be lower cost than the region variants, but if the same cost, might as well just use the region ones, I'd guess
[16:23] Simon Linden: Anyway, it should be in testing next week, so it might hit an RC channel the following one if things go smoothly
[16:23] Mudslinger Ning: *tries to think of a punchline for "What did the chicken llsay accross the road?"
[16:24] Flip Idlemind: What would be the point of llParcelSayTo
[16:24] Simon Linden: humor
[16:24] Moundsa Mayo: A while back there was some discussion on old Jira about a non-email long distance communkiication method. Just wondered if that ever went any further.
[16:24] Flip Idlemind: Of cooourse
[16:24] Sigma Avro: seems llRegionSayTo is noyt yet anounced on lsl wiki
[16:24] Pauline Darkfury: Keep the comms inside the parcel, Flip
[16:24] Pauline Darkfury: e.g. could llParcelSayTo(av, …) and be guaranteed that you wouldn't spam someone outside the parcel
[16:24] Tankmaster Finesmith: " non-email long distance communkiication"... snail mail?
[16:25] Kadah Coba: Prim HTTP-in was good for that
[16:25] Techwolf Lupindo: What is the main difference of llRegionSayTo() and llRegionSay()?
[16:25] Kelly Linden: So, for mono2performance....
[16:25] Simon Linden: We haven't looked at long distance communications
[16:25] DogWomble Dollinger: tankmaster nah i think it's sneakernet
[16:25] Tankmaster Finesmith: lol
[16:25] Kelly Linden: We have a new version up this afternoon that addresses SCR-18
[16:25] JIRA-helper: http://jira.secondlife.com/browse/SCR-18
[#SCR-18] Large many element list slow scripts down
[16:25] Draconis Neurocam: im sure with the new xmpp stuff you could piggyback some communication structure for lsl...
[16:25] Flip Idlemind: Yeah I get it but I don't think you'd need a whole new function, just an if statement
[16:25] Simon Linden: Techwolf - it's sent directly to an object, others can't hear the message
[16:25] Kadah Coba: llSendCarrierPigon()
[16:25] Sigma Avro: Yes Kelly ?
[16:26] Pauline Darkfury: http_in isn't usable without an off-world component, unfortunately. Something which could send a targetted message to (region, object_uuid) would be very handy for certain things
[16:26] Moundsa Mayo: Maybe it's still visible in archived Jira. It drew some thoughtful commentary.
[16:26] Techwolf Lupindo: Simon, oo...I know a lot of products that can use that for better security.
[16:26] Flip Idlemind: "addresses" or "fixes"?
[16:26] Kelly Linden: The problem there was an unknown new function to track the high watermark on mono script memory was forcing a recount of memory way too often.
[16:26] Moundsa Mayo: Yes, Pauline.
[16:26] Draconis Neurocam: ouch kelly
[16:26] Sigma Avro: wich one ?
[16:26] Kelly Linden: We still want to enable tracking that we just can't do it constantly.
[16:26] Draconis Neurocam: right
[16:26] Maestro Linden: (unknown to the public, not unknown to kelly.. I hope ;) )
[16:26] Sigma Avro: ah, ok
[16:27] Kelly Linden: So I replaced the function since no one knew about it yet with a new system that lets you enable and disable memory profiling at significant peformance cost on large scripts.
[16:27] Sigma Avro: next present to sl, hehe
[16:27] Kelly Linden: Yes by unknown I mean we didn't publish that llGetMaxUsedMemory existed
[16:27] Flip Idlemind: Does it?
[16:27] Kelly Linden: Not any more.
[16:28] Pauline Darkfury: Even if it had an annoying delay, llGetActualMemoryUsage() or similar would be wonderful for optimising stuff
[16:28] Sigma Avro: great !
[16:28] Kelly Linden: There are now 3 new functions
[16:28] Flip Idlemind: Go oooon
[16:28] Pauline Darkfury: Maybe even llGetBytecodeSize() and llGetDataSize()?
[16:28] Sigma Avro: the actual is really a great acheivement
[16:28] Kelly Linden: llGetCurrentUsedMemory will report the (surprise) current used memory for mono scripts. It doesn't work for LSL (reports 16k always) and is unencumbered by the legacy behavior that cripples llGetFreeMemory
[16:29] Kadah Coba wants real arrays still :(
[16:29] Draconis Neurocam: and the other two?
[16:29] Sigma Avro: ...but
[16:29] Sigma Avro: ...waht is ...
[16:29] Sigma Avro: ...the meaning of actual ?
[16:29] Kelly Linden: llScriptProfiler which accepts currently the flags PROFILE_NONE and PROFILE_SCRIPT_MEMORY
[16:29] DogWomble Dollinger: kelly, does this mean work is being done that will allow scripters to define memory allocation in mono to something other than 64k?
[16:30] Kelly Linden: And llGetSPMaxUsedMemory which will report the high watermark for used memory while PROFILE_SCRIPT_MEMORY was last (or is currently) active.
[16:30] Kelly Linden: oh, that is llGetSPMaxMemory, I mean.
[16:30] Kelly Linden: Dog: This is a precurser to that, yes.
[16:31] Draconis Neurocam: what do you mean by high watermark exactly?
[16:31] Sigma Avro: yes,n what is actual ?
[16:31] Ardy Lay: highwater mark
[16:31] DogWomble Dollinger: excellent kelly :)
[16:31] Kelly Linden: If a script allocates a few lists and then releases them and then you call llGetCurrentUsedMemory you will get something that doesn't tell you how high you went while you were allocating lists.
[16:31] Mudslinger Ning: will we need noah to rez an ark?
[16:31] Ardy Lay: Flood water leaves a mark on stuff when it goes down. ;-)
[16:31] Simon Linden: It's more like 'high tide'
[16:31] Sigma Avro: exactly, so ok, that is what we wished
[16:32] Draconis Neurocam: ahh
[16:32] Draconis Neurocam: neat
[16:32] Draconis Neurocam: thats cool actual
[16:32] Draconis Neurocam: actually*
[16:32] Sigma Avro: thanks Kelly, really cool
[16:32] Ardy Lay: So th mark is called the highwater mark. Like a peak reading meter or something.
[16:32] Kelly Linden: However if you were profiling while the list work was happening, llGetSPMaxMemory will tell you what the highest amount of memory at one time you were using, while profiling.
[16:32] Pauline Darkfury: Yeah, sounds perfect, both accurate HWM & current
[16:32] Flip Idlemind: Question is how long until we see all that stuff on Agni
[16:33] Draconis Neurocam: will some form of getcurrentusedmemory find its way as a flag into getobjectdetails?
[16:33] Kelly Linden: It isn't perfect - there is still some grey area around mono memory allocation, particularly for lists and strings. But it is the same utility exactly that is used to trigger stack heap collisions in mono scripts.
[16:33] Kelly Linden: draconis: unfortunately no.
[16:33] Draconis Neurocam: darn
[16:33] Kelly Linden: draconis: we can only walk the memory graph of a script that is loaded.
[16:34] Kelly Linden: Dog: I can't answer private IMs while I am talking to everyone.
[16:34] DogWomble Dollinger: ok, no prob
[16:34] Sigma Avro: lol
[16:34] Sigma Avro: I would have a suggestion for chats
[16:34] Kelly Linden: draconis: so that means we can't have one script find out the current used memory of a random other script without first loading it or reverting to the really slow profiling behavior to enable the detailed tracking.
[16:35] Draconis Neurocam: right
[16:35] Draconis Neurocam: i understand that
[16:35] Draconis Neurocam: i dont even want to think about how lsl memory is kept track of then
[16:36] Draconis Neurocam: non mono anyway
[16:36] Sigma Avro: and how often could this command be called, I mean what is the foressen time
[16:36] Sigma Avro: *this function
[16:36] Kelly Linden: LSL actually has a 16k buffer that starts the stack on one end and the heap on the other and they grow to the middle. They really do always use 16k of memory.
[16:37] Draconis Neurocam: interesting
[16:37] Kelly Linden: sigma: for which ones?
[16:37] Sigma Avro: the memory one
[16:37] Sigma Avro: the actual
[16:37] Kelly Linden: The memory profiling will slow down your script by 10-100x on large scripts.
[16:37] Sigma Avro: oh, ok
[16:37] Draconis Neurocam: is some linden going to make official function pages for all three of those, along with regionsayto, when they are available?
[16:37] Maestro Linden: you can call the functions themselves really rapidly though
[16:37] Kelly Linden: Calling the 'current' one you can do as often as you want. It will still be lower precision.
[16:38] Kelly Linden: Yes I will work on some wiki pages for the memory ones, especially since there are more caveats.
[16:38] Sigma Avro: uncertainty principle nicely ilustrated
[16:38] Pauline Darkfury: llGetHeisenburgMemory()? :p
[16:38] Sigma Avro: interesting to test it, will they be first on beta ?
[16:39] Draconis Neurocam: llScriptVelocityAndPosition
[16:39] Simon Linden: Draconis - yes, we'll update the documentation
[16:39] Draconis Neurocam: cool
[16:39] Kelly Linden: Sorry, I should clarify that 'current' statement. Calling llGetCurrentUsedMemory may be slow and use up a bit of your scirpt time, but the profiling tracking is essentially doing that memory call many times per script time slice which is what makes it so slow
[16:39] Kelly Linden: In comparison calling it one is not a big deal.
[16:40] Kelly Linden: once*
[16:40] Maestro Linden: ? - you can try out the mono2performance functions rihgt now on Aditi, if you go to a 'mono2performance 11.04.08.226042' region
[16:40] Kelly Linden: Yes, what maestro says. :)
[16:40] Sigma Avro: ok, thanks MAestro, when doc is avalaible ?
[16:40] Sigma Avro: tks Kelly
[16:41] Pauline Darkfury: Mmm, now we've got accurate-ish memory reporting, how about something like llGetScriptTime() for the CPU side of it?
[16:41] Kelly Linden: Another caveat is that the profiling state will reset to NONE whenever the script is reset, rezed, derezed, crosses a region border.
[16:41] Draconis Neurocam: that sounds fair
[16:41] Sigma Avro: yup, true
[16:41] Kelly Linden: Pauline: the profiling is setup so that we can add things like PROFILE_SCRIPT_CPU later as a possiblity.
[16:41] Pauline Darkfury: cool
[16:41] Draconis Neurocam: usually the only person interested would be the writer, if not it would be opensource anyway
[16:41] Kelly Linden: but those are currently not in the works.
[16:42] Pauline Darkfury: anything which gets perf info out to people other than EMs is a good thing, pretty much. So many people scripting in the dark about what impact their scripts really have an a region
[16:42] Kelly Linden: And the performance hit is bad enough that you really don't want it set for production items
[16:43] Pauline Darkfury: Yeah, I see these things as a development-phase tool for optimising stuff before release
[16:43] Kelly Linden: yup.
[16:43] Sigma Avro: it could even be autoamticaly added at first use of the script, lol
[16:43] Draconis Neurocam: many many people will be excited by this
[16:44] Techwolf Lupindo: Now if only we could get those tool to work on no-mod scripts so us users will know how much that HUD uses....]
[16:44] Kelly Linden: With these changes we find that (at least while profiling is disabled) mono scripts in mono2-performance regions run as fast or faster than their counterparts on agni.
[16:44] Draconis Neurocam: no more suprises hopefully
[16:45] Simon Linden: Thanks Kelly ... we have 15 minutes left, so let's open the table for any questions or topics
[16:46] Pauline Darkfury: SVC-6880, new instance of it reported this week, confirming my observations on it (I think)
[16:46] Pauline Darkfury: Hmm, Jira helper doesn't like commas - SVC-6880
[16:47] Pauline Darkfury: (or it's just not working?)
[16:47] Tankmaster Finesmith: https://jira.secondlife.com/browse/SVC-6880
[16:47] JIRA-helper: [#SVC-6880] Texture settings randomly lost on in-world object for no obvious reason
[16:47] Sigma Avro: Yes, it seems to me that there is a dissimetry when crossing sim borders of different rc regions. Well seems natural, but quite llarge a difference (menawhile, sorry Pauline)
[16:47] Liisa Runo: jira helper remember it been mentioned recently
[16:47] Ardy Lay: Was siad already this hour
[16:47] Kadah Coba: LSL needs regex :3
[16:48] Pauline Darkfury: ahh, ok, didn't know it had memory
[16:48] Draconis Neurocam: on the linden developer side of things, have things been getting better overall?, understandable if you dont feel right answering
[16:48] Maestro Linden: Hm, I looked at Void's report in SVC-6880, but I kind of hit the same wall. I saw that PRIM_PHANTOM was explicitly disabled on his object by something in the sim (maybe a script?), but unfortunately I couldn't find any useful details about what changed the texture on his object..
[16:48] Simon Linden: uh, that one looks tough if we can't figure out how to reproduce it
[16:48] Pauline Darkfury: Yeah, it's a no-repro right now, seems to happen randomly with quite low probability
[16:49] Sigma Avro: rezing time maybe ?
[16:49] Sigma Avro: viewer side ?
[16:49] Pauline Darkfury: I've seen it now when rezzing an object (2nd rez of the same object is fine), shrugged that off initially, but now have seen it on objects just happily sitting in-world
[16:49] Pauline Darkfury: It's changing the object state on the server, take a copy, rez the copy, plywood is still there
[16:50] Sigma Avro: are these 1024 textures ?
[16:50] Pauline Darkfury: Logout, nuke cache, login, plywood still there
[16:50] Pauline Darkfury: 512x512 in my case
[16:51] Simon Linden: I doubt the size is a factor, unless it's a file transfer issue
[16:51] Pauline Darkfury: My hunch is that somehow the prim data is getting corrupted when changes are made via LSL (changing stuff other than textures)
[16:51] JeanMark Nightfire: i have a question, totally unrelated... can anyone tell me how to see invisiprims?
[16:52] Arawn Spitteler: You migh try selecting them
[16:52] Draconis Neurocam: jean, probably the best way is to move your camera so water or avatar mesh is behind them
[16:52] Sigma Avro: happens to me that waves (paticles) take the texture of my ftiings lol
[16:52] Pauline Darkfury: In my case, 99.9% certain the only changes are llSetLinkPrimitiveParamsFast(individual_prim_number, [PRIM_ROT_LOCAL,…]) called approx 1Hz
[16:52] JeanMark Nightfire: mmm they are invisable and dont select
[16:52] Draconis Neurocam: oh, you mean prims with no visible faces then
[16:52] Draconis Neurocam: drag selecting is kinda the only way
[16:52] Sigma Avro: oh, thta does not actually work at this speed indeed, can confoirm
[16:52] Pauline Darkfury: yeah, or rez an alpha texture the far side of the invisiprim
[16:53] Sigma Avro: depends on number of links
[16:53] Kelly Linden: I need to run. Have a nice weekend.
[16:53] Draconis Neurocam: take care kelly
[16:53] Tankmaster Finesmith: you to kelly
[16:53] JeanMark Nightfire: would going into wire frame make them visable?
[16:53] Pauline Darkfury: Take care, Kelly
[16:53] Sigma Avro: by Kelly and thanks a lot, nice we !
[16:53] JeanMark Nightfire: good night kelly
[16:53] Simon Linden: That's a good idea, some of the other rendering modes may help
[16:53] Moundsa Mayo: Thanks for your time, Kelly!
[16:54] Pauline Darkfury: Advanced->Rendering->Info->BBoxes should show them, although cluttered if there's a lot of stuff around
[16:54] JeanMark Nightfire: okies... i will try that, thank you pauline
[16:54] Simon Linden: Nice clock, btw
[16:54] Pauline Darkfury: You kinda have to do a Neo looking at the raw matrix thing to see it properly ;)
[16:55] Ardy Lay: I have some of these out and none have turned to plywood yet that I know of.
[16:55] JeanMark Nightfire: yes, it has caused me some serious grief
[16:55] Simon Linden scrolls back looking for missed questions
[16:56] Ardy Lay: I made it out of a beachball from the library. :-)
[16:56] Liisa Runo: invisprim dont seem to be visible in wireframe mode
[16:56] Simon Linden: haha, that's interesting recycling
[16:56] Sigma Avro: really Liisa ?
[16:56] Pauline Darkfury: Anyway, any suspected prims catching plywooditis, please add to the Jira, seems like it's going to be a tough one to figure out
[16:57] Liisa Runo: yea. look at the prim in front of me
[16:57] Ardy Lay: I wanted to give a full-perm copy to somebody without giving them a full-perm prim with me as creator.
[16:57] JeanMark Nightfire: simon thank you for your help about parcel bonus... it is nice to know there is hope... lol
[16:57] Simon Linden: Some hope with a lot of work :)
[16:57] Sigma Avro: sory am in low connection low graphics
[16:57] Ardy Lay: That causes lots of hatefull IMs if some ass uses it to make replicating litter.
[16:57] JeanMark Nightfire: dont mind the work, just dont want my renters prims to kapoof
[16:57] KiwiAshiri: Clever, Ardy.
[16:57] Techwolf Lupindo: Someone gave me two snapshots, one of what they see and one of what other see. One had default plywood textrues, the other was normal stone textures.
[16:58] Simon Linden: That makes it sound like a viewer updating problem, Techwolf
[16:58] Pauline Darkfury: Yeah, one of my bits of guesswork is that if it's corrupted unexpectedly, the change to plywood won't get pushed to the viewer until something causes the prims to be downloaded again
[16:59] Ardy Lay: Here is another I made with my own prims. :-)
[16:59] Techwolf Lupindo: That what I was thinking, but this was in my deafult download directory, so don't know who or why they was sent to me/group.
[16:59] Ardy Lay: I'll put it on a different time zone.
[17:00] Pauline Darkfury: My version ;)
[17:00] Pauline Darkfury: Mine only does SLT for now
[17:01] Simon Linden: It would be interesting to clear your viewer cache when you see the problem, or maybe clearing it will make it show up more
[17:02] Pauline Darkfury: Yeah, tried completely killing cache over a relog, plywood was still there
[17:02] Draconis Neurocam: what happens to blacklisted textures out of curiousity?
[17:02] Pauline Darkfury: copy in inventory taken before the plywood appeared was fine
[17:02] Pauline Darkfury: fresh copy taken from the world and re-rezzed had plywood
[17:02] Sigma Avro: .. sory, by all, and thanks Simon, Bye Maestro, nice we all
[17:02] Ardy Lay: Have a script read to you the UUIDs of texutres on affected faces.
[17:03] Liisa Runo: griefer cubes with blacklisted textures seem to be gray
[17:03] Simon Linden: I have to get running ... thanks everyone for coming today
[17:03] KiwiAshiri: thanks Simon
[17:03] Pauline Darkfury: Don't think I tried pulling the texture UUIDs via scripts
[17:03] Ardy Lay: Thanks Simon!
[17:03] Moundsa Mayo: Thanks for your time, Simon!
[17:03] Draconis Neurocam: thanks simon
[17:03] Pauline Darkfury: Take care, Simon :)
[17:03] Simon Linden: Take care all, see you around
[17:03] JeanMark Nightfire: thank you simon... have a good evening
|Prev 2011.04.05||Next 2011.04.12|