Scripting User group/Transcripts/2011-03-14

From Second Life Wiki
Jump to: navigation, search

Transcript

[09:06] Draconis Neurocam: hello Kelly
[09:07] Kelly Linden: hello.
[09:07] Jonathan Yap: Hi Kelly
[09:07] Kelly Linden: good morning. Sorry for being late.
[09:07] Liisa Runo: morning
[09:07] Liisa Runo: Kal, buy me RL gas chromatography system
[09:08] Kaluura Boa: Afternoon... Or maybe early evening...
[09:08] Kaluura Boa: =^_^= Sure... No problem...
[09:08] Kelly Linden: everyone have a good weekend?
[09:08] Liisa Runo: yea ^^
[09:08] Draconis Neurocam: yeah, did you?
[09:09] Kelly Linden: enh. I survived.
[09:09] Kelly Linden: ;)
[09:09] Kelly Linden: so today is general topics.
[09:10] Kelly Linden: First a project update: mono2-aware (to make way for new mono version and the stall fix went out to all of agni 2 weeks ago.
[09:11] Jonathan Yap: Is there a date for when it will go out on aditi?
[09:11] Kelly Linden: Working now on getting all the versions on aditi updated so we can begin testing the actual mono2 upgrade there, hopefully later this week. But I wouldn't be surprised if it ended up pushed out to next week.
[09:11] Draconis Neurocam: good to hear its soon at least
[09:12] Kaluura Boa: Bleh... Group chat failure... Can't ring the end of the recess...
[09:13] Jonathan Yap: The xmpp server was running again this morning
[09:13] Kelly Linden: I haven't had a chance to play with that - how has it been working for you?
[09:13] Kaluura Boa: That may explain the "Starting session" and nothing after that
[09:14] Kaluura Boa: Doesn't work well so far...
[09:14] Kelly Linden: heh
[09:14] Jonathan Yap: There were bugs but the basic function was there
[09:15] Kaluura Boa: "Unable to start a chat session... blahblah Go f*** yourself!"
[09:15] Jonathan Yap: I think that's what the headless viewer project is being done for -- to load test the xmpp stuff
[09:15] Kelly Linden: Heh. Well I don't have any further scripting updates today. So open for general Q&A
[09:16] Kelly Linden: We are also using it to load test the server, in combination with some bot-army stuff.
[09:16] Kaluura Boa: Why not llIntegerToHex()? This shouldn't be complicate...
[09:16] Kelly Linden: Not complicated but rendered redundant with TheNextLanguage.
[09:17] Liisa Runo: when we get the next language then?
[09:17] Kaluura Boa: Do you have a confirmed date for the arrival of The Next Language?
[09:17] Kelly Linden: Nope, but eventually. Sometime. Probably. :)
[09:17] Draconis Neurocam: hehe
[09:17] Kaluura Boa: Soom(tm)...
[09:17] Kaluura Boa: Soon*
[09:17] Liisa Runo: untill then, we want new LSL functions to keep us intrested
[09:18] Kelly Linden: I'm very resistant to adding functions that will pollute the namespace / api with functions that are trivial or already covered in more robust languages.
[09:19] Liisa Runo: atleast give us some stuff we desperately need, like the ability to return with LSL
[09:19] Kelly Linden: More interested in functions that specifically relate to interaction with the SL specific systems.
[09:19] Liisa Runo: return objectsÄ
[09:19] Kelly Linden: Yes, like that.
[09:19] Kaluura Boa: The problem with that is that LSL misses those basic things... And TheNextLanguage is not coming any time soon
[09:19] Draconis Neurocam: set return timers, reclaim on estate owned land ,etc
[09:20] Kelly Linden: It is only a few lines of code to do that anyway Kaluura.
[09:20] Kaluura Boa: Wasted memory...
[09:20] Kelly Linden: I have a decision chart I should probably share, I think I just need to remove some names from it.
[09:21] Liisa Runo: speaking of memory. YUou think it owuld be possible to get more script mem before the next language?
[09:21] Kelly Linden: I will see if I can multitask to do that quickly.
[09:21] Liisa Runo: -typos
[09:21] Kelly Linden: Yes Well first 'small scripts' (set a lower upper-bound for mono scripts then limits then bigger scripts.
[09:21] Kelly Linden: then the next language
[09:22] Kaluura Boa: Can we have PRIM_SLICE?
[09:22] Draconis Neurocam: ah yes better control of prim deformations
[09:23] Kelly Linden: I took a quick look at PRIM_SLICE. Adding it as a paramter on the existing shapes is way more difficult than it should be.
[09:23] Kaluura Boa: I also would like all params to be extended up to 0.98 instead of 0.95...
[09:23] Kelly Linden: Would be easier to have the separate PRIM_SLICE parameter but ugly too.
[09:23] Liisa Runo: 0.99*
[09:23] Kaluura Boa: We don't mind ugly... as long as it usable.
[09:24] Liisa Runo: yea, if we cant take the easy way, we rather take hard way than to not anything at all
[09:24] Kelly Linden: I'll take another look at it then.
[09:24] Jonathan Yap: I think the issue is the hard way means x, y, and z don't get worked on for a while
[09:25] Kaluura Boa: There are also a lot llSetLink*( missing functions.
[09:25] Kelly Linden: What do you mean kaluura?
[09:25] Kaluura Boa: llSLPP[F] is supposed to fill the gap but I still miss llLinkTextureOffset( for example
[09:26] Kelly Linden: Ah, the ability to set some individual paramters for llSLPP?
[09:26] Liisa Runo: particle and omega for example still missing from llSLPPFast
[09:26] Kaluura Boa: Yes...
[09:26] Kelly Linden: Yes llSLPPF is still missing some.
[09:26] Kaluura Boa: TargetOmega, I'd like that one!
[09:27] Draconis Neurocam: kelly have you ever thought of an automated system to update the syntax highlighting keywords in the viewer, so if something does change it wont lag behind, for instance we still don't have highlights for PRIM_ NAME, PRIM_DESC
[09:27] Kelly Linden: Yes, I have.
[09:28] Draconis Neurocam: or, even if it wasnt in the viewer, a keyword list file hosted somewhere that could be used to keep external editor syntaxs up to date
[09:28] Kelly Linden: I'd love to have the sim send down a hash or version for a keword file and for the viewer to be able to request the file via cap if it had changed
[09:29] Jonathan Yap: The file now is an xml stored in the viewer tree?
[09:29] Kelly Linden: not quite There are two files (at least one a plain text list of non-function keywords
[09:30] Kelly Linden: And then the function/event keywords are pulled from the LSL compiler directly
[09:30] Kaluura Boa: It was partially if not totally reintegrated in the viewer source... Meaning: Not easy to change on the fly.
[09:30] Draconis Neurocam: and kaluura what about setlinkparticlesystem, why does it need to be in primitiveparamsfast?
[09:30] Kelly Linden: so the viewer would need to change to use a complete LLSD as the highlighting wsource
[09:30] Kaluura Boa: Not my words... It was Liisa's... And the wiki says: http://wiki.secondlife.com/wiki/LlLinkParticleSystem
[09:30] Draconis Neurocam: err liisa nor kaluura sorry
[09:31] Kaluura Boa: (I checked)
[09:31] Liisa Runo: i prefer everything to work trough primparamsFast so i can do the change in one hit and get better sync on stuff
[09:32] Kelly Linden: particle systems are too complex
[09:32] Liisa Runo: but i could be happy with separate function too. anything that helps me take out silly lil child scripts
[09:32] Kaluura Boa: I don't think you can achieve any synchronisation with particles... Too random.
[09:33] Kelly Linden: You shouldn't need child scripts for particles with that function above.
[09:33] Liisa Runo: yea, i would not, but untill we get it we need the child scripts
[09:33] Draconis Neurocam: its out now
[09:33] Kaluura Boa: It's available already
[09:33] Draconis Neurocam: im using it in my staff
[09:33] Liisa Runo: it is? neat
[09:34] Draconis Neurocam: yeah
[09:34] Kelly Linden: :)
[09:34] Liisa Runo: then the omega
[09:34] Kaluura Boa: You should check the list from time to time...
[09:34] Draconis Neurocam: something we cant to is targetomega
[09:34] Draconis Neurocam: yeah
[09:35] Draconis Neurocam: also, and i imagine it isnt entirely your department, but even syntax highlighting on the wiki is lacking
[09:36] Liisa Runo: hmm hmm, also llGetSimulatorAgentKeyList( eould be great improvement towards more efficient scripts
[09:36] Liisa Runo: would*
[09:36] Kelly Linden: I have comments on the jira for that. It isn't as easy as you would think!
[09:36] Kelly Linden: There are interesting script and simulator memory repurcusions.
[09:36] Draconis Neurocam: the easier solution would probably be a feed from the viewer
[09:37] Kaluura Boa: LL refuse to implement RLV in the official viewer...
[09:37] Draconis Neurocam: it wouldnt be rlv
[09:37] Draconis Neurocam: just entering or leaving, and key sent on a channel
[09:38] Kaluura Boa: RLV allows the viewer and script to talk to eachother
[09:38] Kaluura Boa: That's the interesting part.
[09:39] Kaluura Boa: Scripted clothing management might be interesting too...
[09:39] Kelly Linden: Here is the new LSL function decision chart I left with my coworkers when I started my leave last november - it isn't perfect but pretty close to my general thinking http://screencast.com/t/9uLfvTwa
[09:39] Liisa Runo: maybe not as easy as i think, but LL been in the path on improving performance and removing hundreds of flying sensors from the sims would prolly improve the performance a bit
[09:40] Kelly Linden: yeah. Some event that triggered when agents entered or left the region and/or parcel would probably be the better approach.
[09:40] Kelly Linden: though it makes it difficult to get the list on a region that already has people in it. Guess you fall back to sensors in that case.
[09:40] Liisa Runo: yup
[09:41] Liisa Runo: but i bet the sim has the list already, it must have, why cant we call it?
[09:41] Draconis Neurocam: kelly do you mean by lines of sl, the input required, or the overall line count when formatting to be readable isnt an issue
[09:41] Draconis Neurocam: lsl*
[09:41] Kaluura Boa: How does the viewer do for its radar? There must be a way to insert a hook...
[09:41] Kelly Linden: It is just a rough first estimate of how difficult it is to do it in LSL
[09:42] Latif Khalifa: viewer uses CoarseLocationUpdate for "radar"
[09:42] Kelly Linden: the server *has* a list of agents, that isn't the hard part.
[09:42] Jonathan Yap: It sounds like we are dancing a bit around client side scripting
[09:43] Kaluura Boa: We've been dancing for a while already...
[09:43] Kelly Linden: The hard part is that returning a list that could right now be 100 uuids long is significant memory size for scripts, and means that if we ever increased the servers capabilities we may break scripts. While that probably falls in the 'tough love' case we'd really rather just not get into that case at all.
[09:43] Latif Khalifa: client side script would be ideal for this, except CoarseLocationUpdate is flawed
[09:44] Latif Khalifa: 8-bit resolution
[09:44] Jonathan Yap: rather than return a long list how about a data server event response?
[09:44] Kelly Linden: What, you want more than 8bits of resolution for height?
[09:44] Kaluura Boa: Who got that idea to store keys in strings instead of long integers? Sigh...
[09:44] Kelly Linden: It would still be a long list
[09:45] Latif Khalifa: tehKellz, 8 bits gives you only 0-1024m with 4m resolution
[09:45] Kelly Linden: yeah. I know Latif.
[09:45] Draconis Neurocam: it should be 4096 really
[09:45] Latif Khalifa: and you can buld up to 4096... bunch of jiras about it
[09:45] Kelly Linden: I know. I was joking.
[09:45] Kelly Linden: It is horrible behavior for the current world.
[09:46] Latif Khalifa: so even client side would have trouble with it
[09:46] Jonathan Yap: how hard would it be to extend the packet from the server with some higher resolution fields added to the end?
[09:46] Kaluura Boa: AFAIK, the minimap of all the client show ppl underground when they are above 1024m
[09:46] Kaluura Boa: clients*
[09:46] Draconis Neurocam: thats why
[09:46] Kelly Linden: it would be better to replace it with a new packet.
[09:46] Latif Khalifa: it's very easy to add another block... but lindens were worried about the size of that message
[09:47] Kelly Linden: yeah, we send those very frequently
[09:47] Kaluura Boa: Can't the current just be extended with a new value that will be invisible to older clients not aware of it?
[09:47] Kaluura Boa: current block*
[09:48] Latif Khalifa: adding blocks to message template can easily be done in backwards compatible way... i'd just add another byte per z, that would add 1 byte per agent on sim
[09:48] Jonathan Yap: you could add 1 byte which would show what group of ranges to use 0..1023, 1024..2047, etc
[09:48] Kelly Linden: We did that once already to add UUIDs I think.
[09:48] Latif Khalifa: yes
[09:49] Kaluura Boa: OK! Sold! :P
[09:49] Kelly Linden: heh
[09:50] Draconis Neurocam: the viewer should really know about the current state of the server, leaving it unknowing like that is terrible in my opinion
[09:50] Kelly Linden: 10 minute warning.
[09:51] Kaluura Boa: Meaning: Type faster, Liisa! :P
[09:51] Liisa Runo: btw, you know anything about the possible limits to llVolumeDetect? some linden told us that it will be removed from megaprims.
[09:52] Kelly Linden: Falcon or andrew would know more about that.
[09:52] Liisa Runo: currently volumedetect is the only way to make object really phantom, the kind of phantom that dont collide with terrain
[09:52] Kelly Linden: VolumeDetect is really crappy performance wise so it wouldn't surprise me
[09:52] Kaluura Boa: The would break things...
[09:52] Kaluura Boa: That*
[09:52] Latif Khalifa: Falcon says all sorts of things... and then recants ;)
[09:53] Latif Khalifa: Liisa, I think the latest is that it wa't be removed
[09:53] Latif Khalifa: won't
[09:53] Liisa Runo: it is crappy performance wise, but hot air balloon with volumedetect take 0.1ms from the sim while the same thig as normal phantom lag the sim like hell
[09:53] Kelly Linden: I think they are adding more ways to adjust the physics representation of things with mesh.
[09:53] Liisa Runo: anyway, if you dont know about it, lets talke about something else
[09:54] Kelly Linden: ok. :)
[09:54] Kaluura Boa: What is the next function we should see arrive?
[09:54] Kelly Linden: I don't have any in the pipe right now.
[09:55] Liisa Runo holds breath
[09:55] Kelly Linden: I'm currently working on what happens to abandonded land.
[09:55] Latif Khalifa: i got 3 surveys from linden lab last week, whoever made them knows SL well, very good questions. i wrote c# scripting as my nr. priority for enhancing content creation in sl. hopefully others did too :)
[09:55] Kelly Linden: Latif is that the 85 million question one?
[09:55] Draconis Neurocam: i just think its odd to consider replacing lsl when there are no public plans at the moment
[09:55] Latif Khalifa: hope springs eternal :)
[09:55] Latif Khalifa: tehKellz yes
[09:55] Kelly Linden: I couldn't believe the size, but I did give input on the scripting questions.
[09:56] Latif Khalifa: it had a lot of questions, but they were not silly like most surveys are.
[09:56] Latif Khalifa: showed that a lot of thought went into it
[09:56] Kelly Linden: yeah, I just fear it is too long for many to bother finishing. Glad you did.
[09:56] Jonathan Yap: Is there a url for that survey or is it one of those random login ones?
[09:56] Kelly Linden: I don't know how it was distributed
[09:56] Draconis Neurocam: i finished it too, i hope the feedback is actually used
[09:57] Draconis Neurocam: email i believe
[09:57] Latif Khalifa: i got two (and i have two premium accounts so it seems they went only to select group (premuim?)
[09:57] Kaluura Boa: I haven't been surveyed since the last time I refused...
[09:57] Kelly Linden: They were targeting content creators, so I'm not sure how they chose the participants.
[09:57] Latif Khalifa: ah
[09:57] Kaluura Boa: Maybe they choose ppl from MarketPlace
[09:58] Latif Khalifa: could be
[09:58] Kelly Linden: Maybe.
[09:58] Draconis Neurocam: i disliked the word limit on some of write ins, heh
[09:58] Kelly Linden: heh
[09:58] Latif Khalifa: heh, i never run into word limit
[09:59] Draconis Neurocam: i hope you guys actually use the feedback though
[09:59] Latif Khalifa: well hopefully mono2 goes out live, and then it would be even more silly not not have c# scripting than it already is :)
[09:59] Kelly Linden: I'll definitely read it.
[10:00] Latif Khalifa: i mean soon. if the mesh ppl wake up and merge it into their own ode
[10:00] Latif Khalifa: +c
[10:00] Kaluura Boa: Please, no! No more things tied to meshes!
[10:00] Latif Khalifa: well mono2 rollout is now tied to them
[10:00] Kelly Linden: I'm actively working to keep scripting improvements from having a dependency on any other project, mesh included.
[10:01] Liisa Runo: yea, people will hate mesh even before it arrive if mesh is to blame for not getting megaprims and stuff that has nothing to do with mesh
[10:01] Kaluura Boa: Right now, meshes are blocking a lot of things
[10:01] Latif Khalifa: tehKellz, then need -aware patch since half of aditi is mesh
[10:02] Kelly Linden: yeah, we have two approaches to fix it. They are working on getting their dev branch suitable for aditi deploy, and on friday I merged the version already on aditi with the mono2-aware. Hopefully one of those can be deployed shortly.
[10:02] Latif Khalifa: fingress crossed :)
[10:03] Kelly Linden: Ok! Thanks for coming. I'll figure out how to get this on a calendar.

Generated with SLog Wikifier