Simulator User Group/Transcripts/2011.11.22

From Second Life Wiki
Jump to: navigation, search


Prev 2011.11.18 Next 2011.11.29

List of Speakers

Andrew Linden Falcon Linden Fancy Detector
Flip Idlemind Jonathan Yap JoyofRLC Acker
Kadah Coba Kelly Linden Kitto Flora
Leonel Iceghost Qie Niangao Renae Daines
Sahkolihaa Contepomi Simon Linden Squirrel Wood
Vernes Veranes Vincent Nacon


[12:01] Sahkolihaa Contepomi: hey Andrew.

[12:01] Andrew Linden: Hello

[12:01] Andrew Linden: Sorry I couldn't attend last Friday's meeting. I had to depart early.

[12:01] Andrew Linden: I'll try to publish those transcripts today.

[12:02] Andrew Linden: There will be no User Group meeting this Friday.

[12:02] Andrew Linden: All of us Linden's will be on vacation on Friday.

[12:03] Andrew Linden: So the news...

[12:04] Andrew Linden: I did send an email out for an update on enabling return of encroaching objects on mainland, but haven't received a reply yet.

[12:04] Kadah Coba: I don't remember seeing such an email...

[12:04] Andrew Linden: I also started to talk to others about the idea of exposing the world map tiles via more predictable locations (via http)

[12:05] Andrew Linden: however I realized I've forgotten some of the details of how the system currently works

[12:05] Andrew Linden: and was therefore having trouble explaining to the others how it should be *different*

[12:06] Andrew Linden: so I was hoping someone here could refresh my memory on how the tiles are currently available and what would be the most ideal way to access them.

[12:06] Andrew Linden: Finally, I'll mention now before I forget...

[12:07] Andrew Linden: The region Sardar was eventually moved to LeTigre last week, but they requested that it move back to Magnum yesterday.

[12:07] Andrew Linden: I hear that the weapons were breaking in new ways. I was hoping someone who knows more about it might be here to give me some more details.

[12:08] Andrew Linden: end of news and my own agenda items

[12:08] Andrew Linden: Regarding the map tiles...

[12:08] Andrew Linden: There were requests to make the map tiles downloadable via more predictable names, something like:

[12:09] Andrew Linden: ?

[12:09] Andrew Linden: where (gridx, gridy) = location of region in the grid

[12:10] Andrew Linden: and lod would be the level-of-detail of the tile

[12:10] Qie Niangao: so... I thought the issue was with the in-world textures corresponding to the maps, currently accessed through an external server. (or maybe that was a different issue?)

[12:10] Kadah Coba: I think that would work, but I have no use for such a thing :P

[12:11] Qie Niangao: AFAIK, the MapAPI already supplies jpgs at findable locations, via http

[12:11] Andrew Linden: Oh hrm. What that the issue? The desire for the in-world map textures to be guessable assets?

[12:11] Fancy Detector: Kelly Linden has arrived!

[12:12] Leonel Iceghost: any way you can make it an uuid to use from scripts?

[12:12] Flip Idlemind: I'd like that

[12:12] Simon Linden:

[12:12] Simon Linden: (fwiw)

[12:12] Andrew Linden: well, suppose we designated region_id.texture as the location of that region's map tile...

[12:13] Andrew Linden: and suppose we updated that texture in the asset system

[12:13] Andrew Linden: that is, instead of just writing once we updated it dialy or on some schedule

[12:13] Andrew Linden: SL's texture asset system pipeline is designed to be cached at several stages

[12:14] Andrew Linden: so if you looked at that texture on Tuesday

[12:14] Andrew Linden: and they also looked on Wed or Thurs

[12:14] Andrew Linden: then the cache may not have updated

[12:14] Andrew Linden: and it is possible for that texture to be cached for a week or more

[12:15] Andrew Linden: depending on how it is used, and how well the variosu caches along the way are working

[12:15] Andrew Linden: That is, if you viewed that texture in your viewer daily... it may be that your viewer cache would keep it around indefinitely.

[12:16] Andrew Linden: oh hey, Kitto Flora is here. Long time no see.

[12:16] Andrew Linden: I think Simon had some news about SVC-22.

[12:16] JIRA-helper:

[#SVC-22] Vehicles crossing region borders aren't always treated as vehicles and can get incorrectly returned if the destination parcel is no-entry or parcel-full

[12:16] Vincent Nacon: no more sailors?

[12:16] Vincent Nacon acrosses his fingers

[12:16] Kitto Flora came to ask about KeyframeAnimation

[12:16] Jonathan Yap: Andrew, I have an IRC link to jira updates and have been going nutty seeing so many comments about svc-7459. Could you look into that one (or, if you already are, mark the jira accordingly) please?

[12:16] JIRA-helper:

[#SVC-7459] Drop In Traffic Calculations

[12:17] Simon Linden: well, there's a claim in SVC-22 that it's fixed

[12:17] Simon Linden: The LeTigre branch has some new object-region crossing code that checks permissions earlier than it used to, so I have guarded hopes that it is indeed fixed

[12:18] Kitto Flora thinks about who may see the SVC22 problem.

[12:18] Simon Linden: But given the legendary status of that bug, I hesitate to claim that ... so if anyone feels like testing and seeing if that's true, I'm interested in what you find

[12:19] Andrew Linden: I'm not currently looking into 7459, but I saw some internal email traffic about it, so I believe there is some attention being paid.

[12:19] Kitto Flora: I do see vehicle loss due to 'parcel full' occasionally

[12:20] Simon Linden: The new code should affect permissions, not neccessarily limits due to prim counts

[12:20] Jonathan Yap: Maybe 7459 could be marked as acknowleged to quiet people down a bit

[12:21] Simon Linden: Good point, Jonathan

[12:21] Andrew Linden: That is good advice Jonathan. I'll try to find out more internally and find someone to post there.

[12:21] Jonathan Yap: Thank you Simon and Andrew

[12:21] Kitto Flora: In all my apps, permissions are always correct - that is I ensure that the next parcel is always accessible. And it is.

[12:22] Kitto Flora: How anyone expects parvcel prims limits to be bypassed for a 'vehicle' that has no rider - beats me.

[12:23] Flip Idlemind: Crossing regions while sitting on a prim works by unsitting you, moving you and the object to the other region, and then re-sitting you, right?

[12:23] Simon Linden: Right, that's basically what happens under the hood

[12:24] Flip Idlemind: So, the auto return could kick in during that brief moment when you're not technically sitting on the vehicle?

[12:25] Andrew Linden: I thought we checked for pending re-seat of avatar in the auto-return code, but I'd have to look at the code to verify.

[12:25] Simon Linden: Right, that's how I'd expect that problem to occur

[12:26] Simon Linden: But I'd have to look at the code as well. Of course, it's all _supposed_ to work right :P

[12:26] Andrew Linden: Kitto, you had a question about keyframe animation?

[12:27] Kitto Flora: Yes - is what we have now for that function 'how its going to be" or can we expect improvements?

[12:28] Andrew Linden: Oh btw, we do have someone working on making avatar + vehicle region crossings atomic, which would eliminate a parcel autoreturn race condition if it existed.

[12:28] Andrew Linden: Kitto, perhaps you could summarize the way it is now for me? I'm not too familiar with the script API for keyframe animation.

[12:29] Andrew Linden: Also, what you would like to see changed/added?

[12:29] Kitto Flora: Er - thats a long explanation...

[12:29] Flip Idlemind: I remember Falcon (I think) saying it was designed so that making improvements to it would be easy

[12:29] Kitto Flora: It takes a list of position,rotation,time,......

[12:29] Andrew Linden: s/easy/possible/

[12:29] Vincent Nacon: position by region instead of local

[12:30] Kitto Flora: and does a llSetPos() type move on the object , linearly moving between the waypoints in the requested time

[12:30] Andrew Linden: Vincent, that region position statement was something for the wishlist right?

[12:30] Meeter: Timecheck : User Group is half over

[12:30] Kitto Flora: The current problem is - it fails at border crossings!

[12:30] Vincent Nacon: yeah, was brought up on the wiki page in talk

[12:30] Kitto Flora: (Dej vu??)

[12:31] Kitto Flora: The object will stop mid list, at 2M outside the region its in, and not make any border crossing

[12:31] Kitto Flora: Essentailly, I suppose - Keyfarem Anim. fails to push an object into the next region.

[12:32] Andrew Linden: Ok the region crossing problem should definitely be fixed. I thought it was working.

[12:32] Fancy Detector: Falcon Linden has arrived!

[12:32] Vincent Nacon: maybe Falcon can answer that

[12:33] Kitto Flora: Nah - have to detect failure to cross then hammer the object sith multiple llSetPus() to get it into next region - sometimes it takes 3 meters of shove to get it into next region

[12:33] Andrew Linden: Falcon. Kitto Flora says keyframe animation motion fails at region crossings. Is that a known issue?

[12:33] Kitto Flora: IE it dont trasition until 5M outside the old region

[12:34] Falcon Linden: Ahoy folks

[12:34] Vincent Nacon: heya

[12:34] JoyofRLC Acker: Ahoy! :)

[12:34] Falcon Linden: Thought I'd swing by today. I was wondering if anybody's had a chance to try out llSetKeyframedMotion on some interesting builds

[12:34] Falcon Linden: or test the new limits in llCastRay

[12:35] Andrew Linden: Falcon, scroll back for a question about keyframe animation failing at region crossings.

[12:35] Jonathan Yap: He might not be able to scroll back

[12:35] Vincent Nacon: he wasn't in chat range at the time

[12:35] Vincent Nacon: [12:33] Andrew Linden: Falcon. Kitto Flora says keyframe animation motion fails at region crossings. Is that a known issue?

[12:35] Falcon Linden: it shouldn't and didn't fail in any of my or maestro's tests

[12:35] Kitto Flora: llSetKeyframedMotion - I have vehicle that runs around the SLRR rail tracks - works well except border crossings all fail - or >90% fail. It always needs ashove.

[12:36] Falcon Linden: that's strange. Can you repro it with a KFM_PING_PONG animation across a single border?

[12:36] Kitto Flora: Dont know. I can try

[12:37] Flip Idlemind: Quick question, is keyframe motion everywhere now, or still on an RC?

[12:37] Falcon Linden: cool. Create the simplest repro you can, open a jira with it, and Maestro will check it out

[12:37] Falcon Linden: FLIP: everywhere

[12:37] Flip Idlemind: Wooty

[12:37] Leonel Iceghost: will it work in huds someday?

[12:38] Falcon Linden: Also happy to see that there don't appear to be any physics bugs introduced by the Havok 2011.2 upgrade. W00t!

[12:38] Falcon Linden: Leonel: not likely, it's intended for interacting with the physics system

[12:38] Falcon Linden: Fundamentally, it's tied to an object's rigid body and attachments don't have one

[12:39] Andrew Linden: That is, attachments don't have a collision shape in the physics engine.

[12:39] Leonel Iceghost: something with that for huds would be amazing

[12:39] Leonel Iceghost: like that*

[12:39] Falcon Linden: so many things would be amazing :)

[12:39] Leonel Iceghost: algo it wouldn't hurt framed color changing and other stuff :P

[12:39] Leonel Iceghost: also

[12:40] Falcon Linden: I vote for not providing that for HUDs and instead providing a nice API for HTML 5 huds

[12:40] Leonel Iceghost: html is blocked by firewalls

[12:40] Leonel Iceghost: first make it so the ip is hidden, so we can use it freely

[12:40] Kitto Flora: Play AngryBirds on your HUD?

[12:41] Leonel Iceghost: firewall as firestorm firewall I mean

[12:41] Falcon Linden: There is no actual dev work going on on that idea as far as I know, btw

[12:42] Simon Linden: You've seen the demo HTML hud, right?

[12:42] Simon Linden:

[12:42] Leonel Iceghost: yes, it asks an ugly permission to allow or deny in 3rd party viewers

[12:42] Vincent Nacon: what about attachment?

[12:43] Falcon Linden: Simon: yeah, I did

[12:43] Simon Linden: You should bug some 3rd party viewer developers :)

[12:43] Falcon Linden: Vincent: what about attachment?

[12:43] Vincent Nacon: instead of hud?

[12:43] Leonel Iceghost: they do that because otherwise people collect ips

[12:43] Leonel Iceghost: to ban

[12:43] Leonel Iceghost: from states

[12:43] Andrew Linden: Actually I think a keyframed HUD/attachment would be a nice feature. I'm willing to ponder how it might be done, but I can think of many things I'd like to fix before tackling that.

[12:44] Leonel Iceghost: is there any way the ip of the user could be hidden for the hud?

[12:44] Vincent Nacon: without physic, of course

[12:44] Vincent Nacon: mainly for display use

[12:44] Leonel Iceghost: a new event

[12:44] Leonel Iceghost: or something

[12:45] Andrew Linden: Ah, Leonel is asking for IP-address obfuscation for HTML huds

[12:45] Falcon Linden: Vincent: right. Yeah, I still think a generalized HUD/UI system is a better use of dev resources, though it would use more of them...

[12:46] Leonel Iceghost: you could show the IP of the server, so 3rd party viewers would allow free use of http huds

[12:46] Vincent Nacon: oh I don't meant for hud

[12:46] JoyofRLC Acker: may i ask a question ?

[12:46] Andrew Linden: Ask JoyofRLC.

[12:46] Vincent Nacon: attachment, like your falcon bird could move around smooth motion

[12:47] Falcon Linden: oh, I see

[12:47] Falcon Linden: yeah, that would be nice.

[12:47] Andrew Linden: Regarding hidden viewer IP addresses... LL would have to run some sort of http proxy service. A bick stack of proxies.

[12:47] JoyofRLC Acker: ... my apologies i think i arrived in the moddle of something, this talk about HUDs - it sounds intriguing; is there somewhere i can see what its about, i looked at the demo quickly but i thnk id need a bit of an itnroduction

[12:47] Vincent Nacon: but of course... I know some people gonna want that on HUD as well but I'd agree about HTML5

[12:47] Leonel Iceghost: why a proxy, just intercept the request to the ip flag

[12:48] Vernes Veranes: traffic would still need to be rerouted.

[12:48] Jonathan Yap: How is this html5 suggestion different than using Media on a Prim (with maybe an updated to qtwebkit)?

[12:48] Vincent Nacon: less resource on SL's server

[12:49] Falcon Linden: jonathan: media on a prim doesn't allow, e.g., transparent backgrounds. It also doesn't provide a mechanism for interacting with the world

[12:50] Falcon Linden: the kind of system I'm envisioning would offer some way of interacting with SL from inside the UI. But I haven't spent much time thinking about it

[12:50] Andrew Linden: A summary for JoyofRLC: we were talking about keyframed animation, someone asked if we could have it for HUD's/attachments, Falcon replied he'd rather see HTML-5 on HUD instead. Simon pointed at the scripted HTML-HUD demo mini-map example.

[12:50] Vincent Nacon: just don't limit it to Flash, like what BlueMars did.... muhaha!

[12:51] JoyofRLC Acker: Thanks andrew.

[12:52] Qie Niangao: So... returning to Maps for a moment, but this time specifically about the MapAPI ... (I mumbled about this a few meetings back) ... will there be a problem next year, when Google starts charging to use their Map API? The API-generated SL maps have a Google copyright, even though obviously not Google's geo data, so... not sure if LL will suddenly have a big bill to pay Google each month or not.

[12:52] Kelly Linden: Media on a prim from LSL scripts has decent interaction with the world (back to the LSL script). The transparency part is a problem, but I don't think it is unsolvable.

[12:52] Flip Idlemind: Google needs more money of course. Just like Microsoft does

[12:53] Andrew Linden: Qie, that is a good question. Dunno the answer. I"ll have to ask some others in LL.

[12:53] Squirrel Wood: Everyone needs more, more, more. That's what its all about :p

[12:53] Qie Niangao: Thanks. I appreciate it.

[12:53] Leonel Iceghost: they don't do it for themselves, they need money for they grand-grand-grand sons

[12:54] Leonel Iceghost: college

[12:54] Laggi Meeroo whispers: Simon Linden will you please pick me up and snuggle me?

[12:55] Leonel Iceghost: html would be awersome if we can just use as we want for everybody, without perms or things like that

[12:55] Meeter: Timecheck : User Group is almost over

[12:55] Leonel Iceghost: for webs hosted enterily in LL, I don't see why it needs the ip

[12:56] Vincent Nacon: making you JIRA feature for the Region/Local flag, Falcon

[12:57] Flip Idlemind: So, Kelly, I remember you mentioning something about a really short llSleep inside while() loops and whatnot being much nicer to the sim

[12:57] Kelly Linden: yup

[12:57] Flip Idlemind: Is that still up-coming, or is it currently true

[12:58] Falcon Linden: Vincent: no, not that. I meant for the KFM_SHIFT stuff

[12:58] Kelly Linden: A while(1) llSleep(0.001); should be nearly as performant but way nicer on the sim than a while(1) ; especially if you are waiting for something that won't change while the script runs.

[12:58] Falcon Linden: Vincent: I'm not adding a region/local flag--it would complicate serialization which is already kind of touchy

[12:58] Kelly Linden: Flip: I think that is on letigre.

[12:58] Squirrel Wood: heh. I still have mono scripts that have a scripted prim color chane that never gets done.

[12:58] Vincent Nacon: um... that's what it was about, Falcon...

[12:58] Squirrel Wood: change

[12:59] Vincent Nacon: what did you think it was?

[12:59] Leonel Iceghost: Kelly you did it so llSleep(0.001) didn't skip a frame, and then you undid that?

[13:00] Meeter: Thank you for coming to the Server User Group

[13:00] Falcon Linden: vincent: well, there were two suggestions in that thread. One was about providing a flag for local/absolute coordinates. The other was about providing two extra parameters to let you shift or rotate a whole list

[13:01] Kelly Linden: Leonel: I think I did fix a bug in that.

[13:01] Falcon Linden: Looks like OH are over and I need to hassle someone about something, so I've gotta run

[13:01] Vincent Nacon: seeya

[13:01] Flip Idlemind: Have fun

[13:01] Falcon Linden: Cheers all! And, if you're in a country that celebrates it, enjoy your Thanksgiving

[13:01] Leonel Iceghost: what happened.. people didn't want the boost in velocity?

[13:01] Kelly Linden: o/

[13:01] Falcon Linden: (And if you're not, enjoy the day anyway?)

[13:02] Squirrel Wood: Enjoy!

[13:02] Sahkolihaa Contepomi: See you Falcon, Kelly, Simon & Andrew.

[13:02] Squirrel Wood: ...

[13:02] Andrew Linden: Thanks for coming everyone.

[13:02] Renae Daines: And if not, use it as an excuse for a nice meal =p

[13:02] Squirrel Wood: lag chat

[13:02] Qie Niangao: Thanks all... have fun!


Prev 2011.11.18 Next 2011.11.29