User:Ama Omega/archive/Office Hours/2010-07-19

From Second Life Wiki
Jump to: navigation, search

List of Attendees


[09:03] Ardy Lay: Hello Kelly
[09:03] Kelly Linden: Hello
[09:03] Kelly Linden: How is everyone? Have a good weekend?
[09:03] Kitto Flora: There was a weekend?
[09:03] Kelly Linden: Yeah, I missed mine too.
[09:04] Liisa Runo: weekend was good here, as good as any other days, life rules
[09:04] Kelly Linden: :)
[09:05] Kelly Linden: Does anyone have a chat log from last week? I didn't think to save one to put it up.
[09:05] Ardy Lay: I think I missed last week. (looks)
[09:06] Liisa Runo: i think i will crash if i try to open the open chat log, it is 40gigabytes or something
[09:06] Kelly Linden: haha
[09:06] Kelly Linden: ok, I'll see if I can sift it out as well.
[09:06] Kelly Linden: Yet again I don't have a good agenda.
[09:07] Kelly Linden: I looked into llSetLinkPrimitiveParamsFast being .... not fast .... last week.
[09:07] Ardy Lay: Liisa, use grep
[09:08] Jonno Stromfield: I missed it. Was there a conclusion as to how to get multiple child prim movements synchronised?
[09:08] Kelly Linden: Turns out it is specific to PRIM_SIZE and PRIM_POSITION on child prims, which trigger a check against the link rules.
[09:08] Liisa Runo: will it be better _when_ we get the link rules simplified?
[09:08] Kelly Linden: Looking at the link algorithm it is complex, but my first guesses at what was really slow didn't quite pan out. I should be able to track it down this week though.
[09:09] Kelly Linden: That would be the hope
[09:09] Liisa Runo: cool
[09:10] Liisa Runo: anyway, i never expected it be capable of cynch setting, for that we need something like llSetLinkListParams
[09:10] Liisa Runo: synch*
[09:10] Jonno Stromfield: can live in hope
[09:10] Kelly Linden: I'd hope we can get it up to near the speed of the other params - like PRIM_TEXT for example
[09:10] Kelly Linden: Which takes 1-2s for 100 prims
[09:11] Jonno Stromfield: that would be 2-3 prims per sim frame?
[09:11] Rex Cronon: hello everybody
[09:12] Liisa Runo: /me breaks the silence: got some idea for you new office yet?
[09:12] Liisa Runo: your*
[09:12] Kelly Linden: closer to 1-2 I think but somewhere around there.
[09:12] Kelly Linden: 1 / frame is 2s for 90 prims
[09:13] Kelly Linden: I have a plot of land just up the hill over there
[09:13] Kelly Linden: it is just a cube farm though.
[09:13] Kelly Linden: I'll get something set up and change the wiki
[09:14] Kelly Linden: Anyone have any questions or topics?
[09:14] Kelly Linden: Strong opinions on the link rules?
[09:14] Rex Cronon: when is textbox going to work?
[09:15] Liisa Runo: only strong opinion is that they need to be simplified, and everybody agree on that
[09:15] Kitto Flora: I wondered what the point of the link rules are
[09:15] Kaluura Boa: When will we have PRIM_SLICE?
[09:15] Jonno Stromfield: which link rules question are we talking about? Avatar link distance?
[09:15] Kelly Linden: Textbox works on the server, I dunno when the viewer will support it.
[09:15] Ann Otoole: and the commandment of thou shalt not break existing content lol
[09:16] Rex Cronon: yes. link rules. why can't u move a prim that is too far in a linked set closer to the root prim?
[09:16] Rex Cronon: i understand not moving it away, but why not allow it to get closer?
[09:17] Kelly Linden: Link rules have 2 main benefits I've been able to determine. They make it more difficult to accidentally link your necklace to their house. The other is for physics performance
[09:17] Kelly Linden: Rex: the rules don't center around the root prim, so we can't do a simple check like that, we have to just run the whole algorithm again - or at least that is what we do now
[09:17] Liisa Runo: if someone has scripted something that rely on the messy link rules, that toy really deserves to go boken and get updated. The current link behavior just cause problems, we need it simplified.
[09:17] Kitto Flora: Accident linking isn't worth much
[09:18] Liisa Runo: broken*
[09:18] Kelly Linden: Right, I'm not a huge fan of keeping them. But the 'they deserve it' rule only flies until it is your toy or giant starship that breaks
[09:18] Kitto Flora: Physics - Humm - dunno how H. deals with islands
[09:18] Ann Otoole: i really wish there was an easier way to reliably control link order
[09:19] Liisa Runo: i can update my toys, hell i even re-buy everything rather than stay in stone age
[09:19] Ann Otoole: so if you have to add or remove a prim it doesn't change all of them lol
[09:19] Rex Cronon: why not use the root prim as the center point?
[09:19] Kelly Linden: havok has 2 phases a broad and narrow phase. If you have two small objects far apart, everything in between will be forced into the more expensive narrow phase. We haven't tested to see what the real impact in practice of that is.
[09:20] Opensource Obscure: not sure I understood this phases thing
[09:20] Kelly Linden: Rex: it just isn't what the current algorithm does and it would break a LOT
[09:21] Rex Cronon: can u have an example that would break?
[09:21] Kelly Linden: Rex: of existing content?
[09:21] Rex Cronon: yes
[09:21] Kelly Linden: your chest piece
[09:21] Kelly Linden: your hat
[09:21] Kelly Linden: your legs
[09:22] Kelly Linden: maybe not your shoes
[09:22] Rex Cronon: my hat is only 2 sculpties. very hard to break
[09:22] Kelly Linden: anything where the root prim isn't at the center of the object
[09:22] Rex Cronon: i have quite low prim count
[09:22] Kelly Linden: it has a root prim little ball above it
[09:22] Rex Cronon: the little ball is an addon
[09:22] Rex Cronon: :)
[09:23] Kelly Linden: anything with an off center root prim potentially would break is the point.
[09:23] Kelly Linden: and since that has never mattered before, lots of content has off center root prims
[09:23] Kelly Linden: your avatar is low prim count?
[09:23] Ann Otoole: well that would be a bummer to have to go back and add little spheres in the middle on items we forgot to put them on :P
[09:24] Kelly Linden: yeah, so I don't think that is a good solution
[09:24] Jonno Stromfield: not to mention costly
[09:24] Kaluura Boa: What's the problem if all the prims of an object are on one side of the root instead of all around?
[09:24] Rex Cronon: if u say nothing can be further than 64m from the root prim. that wouldn't break anything. IMO:)
[09:24] Kitto Flora: All those offshore islands?
[09:24] Ann Otoole: i always found it tends to make attachment positioning a little better when you have that root prim set and positioned right
[09:24] Kelly Linden: Rex: well thats just gone off in a totally different direction.
[09:25] Rex Cronon: u said simpler linking rules. right:)
[09:25] Kelly Linden: 64m from root prim gives you a 128m or half sim diameter range
[09:25] Kelly Linden: Why bother with limits at all at that point? :)
[09:25] Rex Cronon: and considering that now the max distance is 54m, then nothing will get broken:)
[09:25] Kaluura Boa: That's good... I like the idea... And a lot of builders will love it too!
[09:25] Ardy Lay: He wants to make a REALLY BIG gun. :-)
[09:26] Rex Cronon: no
[09:26] Rex Cronon: i wan't a big ship:)
[09:26] Rex Cronon: or a castle that doesn't come in parts:)
[09:26] Kelly Linden: We are more likely to continue with a bounding sphere aproach - it reduces the complexity by not making link order matter
[09:27] Liisa Runo: and sizes of prims in the set
[09:27] Rex Cronon: to me it lookes like what i said would be way, way more simpler than the current linking rules, or any rules that have to use a bounding sphere aproach:)
[09:28] Ann Otoole: yesss larger reptilian pirate sssshipssss are needed
[09:29] Jonno Stromfield: Can I ask whether you would consider changing linked avatar distance rules?
[09:29] Liisa Runo: alright, what else is on the scope? any neat stuff on the way?
[09:29] Rex Cronon: it would be a very KISSable solution:)
[09:29] Kelly Linden: Jonno: you mean for attachments?
[09:29] Jonno Stromfield: 300m as a sit target but less if you move them
[09:29] Jonno Stromfield: as a link
[09:30] Rex Cronon: at least 256m:)
[09:30] Jonno Stromfield: about 70 isn't it?
[09:30] Liisa Runo: yea, for a while there was no limit on how far you can offset agent from the prim it is sitting. this was awesome for many toys, local teleporters for example.
[09:30] Kelly Linden: Liisa: nothing new. Still working on getting the mono-sim-freeze fixes into the main code base (slow but steady progress), I think falcon wants a new revision of llCastRay up in the next week or two
[09:30] Kelly Linden: I don't know of any plans to change that.
[09:31] Jonno Stromfield: okay.
[09:31] Jonno Stromfield: someone mentioned PRIM_SLICE.
[09:31] Kaluura Boa: Yeah...
[09:31] Liisa Runo: well, the current limit is lil silly since sit target allows 300m. i believe it will be smarter some day
[09:31] Opensource Obscure: re: mono-sim-freeze, that's good news..
[09:31] Kaluura Boa: We're waiting for it since slice was officially supported on all prims
[09:32] Kelly Linden: I'll look into prim_slice
[09:32] Liisa Runo: thank you :)
[09:32] Kitto Flora: hole offset would be more useful
[09:32] Rex Cronon: how would prim_slice work?
[09:32] Liisa Runo: prim slice is what you can do with client, but not with script, yet
[09:32] Jonno Stromfield: It's just the same slice as you get in the viewer
[09:32] Kaluura Boa: Some ppl want to add a param to all PRIM_TYPEs... Not likely...
[09:33] Kelly Linden: part of the trick is that it really should be part of the PRIM_TYPE .... right that is tough.
[09:33] Kaluura Boa: Just a PRIM_SLICE, <start, end, 0> would make a lot of ppl happy
[09:33] Jonno Stromfield: if you can't change some scripts that are already sliced without losing the slice
[09:34] Jonno Stromfield: even a don't override the slice would be good
[09:34] Kelly Linden: Probably can't change the existing behavior for the existing constants
[09:34] Jonno Stromfield: sorry I said scripts when I meant prims
[09:35] Jonno Stromfield: not suggesting that, Kelly
[09:35] Jonno Stromfield: I'm suggesting that SLICE is not reset to 0,1
[09:36] Kelly Linden: When is it reset?
[09:37] Jonno Stromfield: If I cut a box in script, it resets the slice
[09:37] Kaluura Boa: Any time you use PRIM_TYPE, even if you don't actually change of type
[09:37] Liisa Runo: nowdays when prim is sliced with viewer, and then you set other prim params with script that affect the shape, the slice is reset
[09:37] Jonno Stromfield: If I cut it in the client tool it's fine
[09:37] Kelly Linden: Right, so we probably can't change the behavior of the existing constants
[09:38] Kelly Linden: So those will continue to reset the slice, and you'd have to set it with the new parameter
[09:38] Liisa Runo: the current behavior is clearly bugged, and people should realise that if they rely on bugged behavior, they need to update theri scripts some day
[09:38] Jonno Stromfield: You already have, kelly and you broke content
[09:38] Kelly Linden: They should, they often don't.
[09:38] Kelly Linden: :O
[09:39] Jonno Stromfield: I'm just asking that you put it back until you fix it properly
[09:40] Kaluura Boa: Another proposed solution in the Jira is to have all new prim types with slice correctly added in them.
[09:40] Jonno Stromfield: when i say you obviously i don't mean you personally!
[09:40] Kelly Linden: What is the jira #?
[09:40] Kaluura Boa: Good question...
[09:40] Jonno Stromfield: One of Strife's I think
[09:41] Kaluura Boa: SVC-5419
[09:41] Kelly Linden: Thanks
[09:41] Liisa Runo:
[09:42] Kaluura Boa: SVC-5485 is a good one too
[09:42] Kaluura Boa:
[09:42] Jonno Stromfield: that was the one I was looking for, Kal. SVC-5485
[09:43] Kelly Linden: I'll have a look through those
[09:43] Jonno Stromfield: thanks
[09:44] Kaluura Boa: Thanks... I don't even remember when slice was allowed and we're still waiting for the script part
[09:44] Liisa Runo: yea, thanks, llSPPF() is the coolest function ever ❤
[09:44] Liisa Runo: llSLPPF() *
[09:44] Kelly Linden: you are only saying that because llCastRay isn't out yet
[09:45] Liisa Runo: ray is cool, but llSLPPF() will still stay the coolest and most usefull one
[09:45] Liisa Runo: there is only about 5 things you can do with ray, done 4 of them already
[09:45] Kelly Linden: what haven't you done?
[09:46] Kaluura Boa: Talking about llSLPPF()... Would you consider add PRIM_LOCAL_ROT and PRIM_LOCAL_POS?
[09:46] Liisa Runo: NPV car that can roll on prim surface
[09:46] Kaluura Boa: adding*
[09:46] Liisa Runo: i started it, but didnt finish yet
[09:46] Rex Cronon: raycast is not so cool, as long as is available only on beta:(
[09:46] Kelly Linden: Kaluura: andrew was working on a new PRIM_ROT that behaved sanely
[09:46] Kelly Linden: I kept screwing it up for him though. He may not be talking to me any more.
[09:47] Liisa Runo: ^^
[09:47] Jonno Stromfield: Sane rotation? [Checks he is still in SL]
[09:47] Opensource Obscure: (( you're , Jonno ))
[09:47] Kelly Linden: PRIM_POS we want to do too, but we want to do it when we can also get rid of the 10m rule.
[09:48] Liisa Runo: 5min job
[09:48] Rex Cronon: why is it so hard to fix the rotation on prims in a linked set?
[09:48] Kelly Linden: And we may want to wait on changing the link rules until we can allow mega prims which is waiting on 'prim encroachment' enforcement
[09:48] Kaluura Boa: Content breakage... Must be avoided...
[09:48] Rex Cronon: the root prim rotates ok, the children r the ones that r having problems
[09:48] Liisa Runo: we can always have new name for it PRIM_POS2
[09:49] Liisa Runo: no broken content, 5min job to add
[09:49] Kelly Linden: It just turns out there are lots of ways it could work.
[09:49] Rex Cronon: u mean people made content relling on a functions that didn't work right?
[09:49] Jonno Stromfield: loads of it, Rex
[09:50] Kaluura Boa: No, ppl found way around the bugs... That's different
[09:50] Jonno Stromfield: I agree with Lisa, needs a new name.
[09:50] Liisa Runo: child rotation is not sane, but many scripts make them still work, just lil more complex scripts
[09:50] Rex Cronon: relaying*
[09:50] Kelly Linden: No, this time it isn't a case of breaking or not breaking content. It would be a new constant. (PRIM_ROT instead of PRIM_ROTATION)
[09:50] Kaluura Boa: Thanks for the short name!
[09:50] Kelly Linden: right, we couldn't fix PRIM_ROTATION because of everyone working around how broken the current one is.
[09:51] Jonno Stromfield: That's great news.
[09:51] Rex Cronon: come on. bring this new prim_rot here. people r waiting for it:)
[09:51] Qie Niangao: ( I always wonderd how the heck Lex ever figured out the workaround )
[09:51] Kelly Linden: But how do you want PRIM_ROT to work on a child prim? Should it get/set the root relative rotation, or the region relative rotation or the local (delta) rotation?
[09:51] Jonno Stromfield: root relative
[09:52] Rex Cronon: i want child prims to rotate around root
[09:52] Liisa Runo: root
[09:52] Rex Cronon: and to rotate around their own center:)
[09:52] Jonno Stromfield: global would cause issues on attached objects
[09:52] Kaluura Boa: Yep... That's the most useful
[09:53] Kelly Linden: And if you do it to the root?
[09:53] Rex Cronon: it might be possible to make scripts that rotate around a specific pos in the sim
[09:53] Liisa Runo: the current behavior
[09:54] Jonno Stromfield: the root current behaviour is okay
[09:54] Rex Cronon: but the most important is to be able to rotate around root and themselves
[09:54] Kelly Linden: And that is, it rotates the whole object in region frame?
[09:54] Liisa Runo: yes
[09:54] Jonno Stromfield: yes, we can translate between local and root
[09:54] Jonno Stromfield: or attachment
[09:54] Qie Niangao: oh! you're suggesting a *local* rotation *only* for the root... interesting
[09:55] Rex Cronon: root can rotate without problem. u don't need to fixt that
[09:55] Kelly Linden: I suggest nothing. Just making sure I'm clear on how people expect things to work
[09:55] Qie Niangao: well... I mean, it *is* a pain to rotate *only* the root prim without the linkset going with it
[09:56] Jonno Stromfield: That's a good point.
[09:56] Ann Otoole: so make the root a little smaller and simulate the root rotation with a child prim
[09:56] Qie Niangao: true, if you can spare the prim
[09:56] Ann Otoole: whatever is done it must be done with complicated algebraic formula to chase away the less than dedicated :P
[09:57] Liisa Runo: can spare the prim, we got pile of free prims when link limit rules got simplified
[09:57] Jonno Stromfield: If the bahviour differs from PRIM_ROTATION, you would have the choice, I suppose
[09:57] Qie Niangao: honestly, I can't remember ever trying to do that, so... guess I don't much care. :blush: just interesting idea.
[09:58] Liisa Runo: choise yes, but messy, and how often that is needed?
[09:58] Jonno Stromfield: /me shrugs
[09:59] Jonno Stromfield: Perhaps we are just talking about Omega
[09:59] Kelly Linden: I will pass this conversation on to andrew. I know he has other things on his plate as well.
[09:59] Kelly Linden: Any other topics for the 1 min left?
[10:00] Liisa Runo: llLogOffCopybotter();
[10:00] Kelly Linden: I may want to drop these to every other week or maybe alternate with some jira triage of LSL requests.
[10:00] Liisa Runo: :/
[10:00] Liisa Runo: we want you every week
[10:00] Jonno Stromfield: you deliver sane rotation and prim_type and i will consider the minute all yours
[10:01] Kelly Linden: Just not quite enough happening right now.
[10:01] Kelly Linden: haha jonno
[10:01] Kaluura Boa: If you fix SVC-93, you'll be in hero in SL...
[10:01] Liisa Runo: ^^
[10:01] Qie Niangao: don't want to take your time away from things, Kelly, but really appreciate the time you can spend.
[10:01] Ardy Lay: Hmm... JIRA triage of script issues would be good as it allows people that cannot be here to participate.
[10:02] Kelly Linden: llSetRot is going to stay broken, the new behavior will only be in llSLPPF and its kin
[10:02] Rex Cronon: the problem right now is that if root prims is NOT at zero rot, than the child prims can't rotate around their own center:(
[10:02] Rex Cronon: if root prims is NOT*
[10:02] Rex Cronon: prim*
[10:02] Ardy Lay: The work around is to divide the child's desired position by the root's position, I think.
[10:02] Liisa Runo: can, just tricky
[10:02] Kaluura Boa: Nevermind... It will be more than greatly appreciated
[10:02] Kelly Linden: Rex, I fear I'm misinterpretting the exact case you are talking about
[10:02] Ardy Lay: rotation, not position
[10:03] Rex Cronon: i even wrote a jira and provided lsl code to show what the problem is:(
[10:03] Kelly Linden: I have to go, thanks all for coming. I'll see about getting a chat log up this time, unless someone beats me to it.
[10:03] Rex Cronon: tc kelly
[10:03] Jonno Stromfield: thank you - sounds promising
[10:03] Qie Niangao: thanks again Kelly
[10:03] Opensource Obscure: thanks Kelly !
[10:04] Ardy Lay: See you later, Kelly.

Generated with SLog Wikifier