Simulator User Group/Transcripts/2012.08.21

From Second Life Wiki
< Simulator User Group
Revision as of 15:36, 21 August 2012 by Andrew Linden (talk | contribs) (Created page with "Simulator_User_Group {| | Prev 2012.08.17 | Next 2012.08.24 |} == List of Spe…")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search


Prev 2012.08.17 Next 2012.08.24

List of Speakers

Andrew Linden Arcane Laval Ashiri
Baker Linden Draconis Neurocam Kelly Linden
Lomoco Binder Motor Loon NeoBokrug Elytis
Qie Niangao Sahkolihaa Contepomi Simon Linden
Theresa Tennyson Youri Ashton Yuzuru Jewell


[12:02] Baker Linden: Hello everyone

[12:02] Lomoco Binder: Hey hey

[12:02] Arcane Laval: Heyy

[12:02] Motor Loon: ey

[12:02] Andrew Linden: Hello.

[12:03] Yuzuru Jewell: Hello,Andrew.

[12:03] Lomoco Binder: Do we have an agenda today, or can we start bringing up other issues?

[12:04] Andrew Linden: The agenda goes like this:

[12:04] Andrew Linden: news, then open table

[12:04] Youri Ashton: hey andrew and simon

[12:04] Youri Ashton: hey kelly and baker

[12:05] Andrew Linden: For news I've got: SCR-372 is actually fixed in a maintenance project -- not going to RC this week, but sometime after that.

[12:05] JIRA-helper:

[#SCR-372] Invalid force in llApplyRotationalImpulse bug on Magnum (pathfinding)

[12:05] Simon Linden: Hello everyone

[12:06] Andrew Linden: I want to work on a few more maintenance bugs before the next project takes all my time, so I'm going to try to track down that bug that Kitto was talking about last week and maybe a few others.

[12:06] Andrew Linden: fin

[12:07] Simon Linden: Let's see - for releases, the updates are on the forum at

[12:07] Baker Linden: I'm doing a final pass on my WSGI script for larege group management, and will be working on the viewer-side later this week

[12:07] Simon Linden: We rolled the bug fixes that were on Magnum to the main channel, and there will be RC updates tomorrow

[12:08] Yuzuru Jewell: How about

[12:08] JIRA-helper: [#SVC-8124] Excessive "ParcelOverlay reliable" messages sent by regions since last rolling restart (2012-08-08)

[12:08] Simon Linden: The new one going to Magnum is another maintenance project - that is the one following the code already in BlueSTeel and LeTigre

[12:08] Andrew Linden: Yuzuru, that is fixed somewhere... maybe in one of the RC's.

[12:08] Andrew Linden: Lemme look it up...

[12:08] Simon Linden: I think that is fixed on both RC channels as of tomorrow

[12:08] Theresa Tennyson: Oskar said it would be fixed on ALL of them.

[12:08] Yuzuru Jewell: Thank you, Andrew.

[12:09] Simon Linden: We fast-tracked it

[12:09] Andrew Linden: Thanks Simon.

[12:09] Ashiri: Many thanks!

[12:09] Motor Loon: Good work

[12:09] Qie Niangao: Good news indeed.

[12:10] Andrew Linden: Ok, so I guess the table is open.

[12:10] Lomoco Binder: Can we talk about a permissions issue with the contents of objects that are deeded to groups? This seems to have been a problem since '07, and I just verified that it still exists yesterday with the latest official viewer.

[12:10] Andrew Linden: Sure Lomoco. What is the bug?

[12:10] Simon Linden: I'm looking at the bug list and don't see much SVC fixes - SVC-7641 is one, but that's not fully fixed - the message gives you more data to find the offending object

[12:10] JIRA-helper:

[#SVC-7641] Object on land sending excessive messages

[12:11] Lomoco Binder: Scripts and notecards that are in objects deeded to groups are not readable by anyone, even the group owner,

[12:11] Lomoco Binder: and even if they were originally in there with full permissions

[12:11] Sahkolihaa Contepomi: I'd like to mention that during a grief attack in FurNation yesterday, using estate return on 6,000 scripted objects would crash the sim /instantly/.

[12:11] Motor Loon: what kind of "estate return"... topscripts?

[12:11] Qie Niangao: Lomoco, there's a way around that: share the individual assets with the group.

[12:12] Youri Ashton: is what I suggested on beta grid as well, no one wanted to listen to me Koli

[12:12] Sahkolihaa Contepomi: Debug > object return one.

[12:12] Sahkolihaa Contepomi: Top scripts wouldn't work. :/

[12:12] Andrew Linden: Sahkolihaa, I believe that crash bug is fixed in an RC...

[12:12] Sahkolihaa Contepomi: Oh, good to know.

[12:12] Lomoco Binder: I believe last I tried it, that option was greyed out in the properties window,

[12:12] Youri Ashton: I wonder why I have to be proofen right before someone even understands the urgency of things

[12:13] Andrew Linden: Lomoco, I will look on to see if that issue has already been added...

[12:13] Lomoco Binder: There's a similar issue, but it's not the same issue, and I can't get on jira for some reason to comment or create a new one.

[12:14] Lomoco Binder: Has anyone else had trouble logging into Jira?

[12:14] Andrew Linden: I didn't find it.

[12:14] Youri Ashton: Already added it to the Jira 1 or 2 years ago, it got deleted because some resident thought it was not important enough

[12:15] Qie Niangao: there's actually a jira issue about people not being able to login to jira... somewhere

[12:15] Andrew Linden: Oh, I need to expand my search, just in case it was misfiled...

[12:16] Lomoco Binder: Duplicate issue from viewer 1.23:

[12:16] JIRA-helper: [#VWR-1121] can't edit/view notecards or scripts inside a friend's object having full permissions granted

[12:16] Youri Ashton: andrew, the jira I filed about it was deleted last year. Jelous people everywhere unfortuantely

[12:16] Qie Niangao: web-4822 for the jira access issue

[12:16] JIRA-helper:

[#WEB-4822] Unable to access

[12:17] Lomoco Binder: It was reopened under

[12:17] JIRA-helper: [#VWR-28738] can't edit/view notecards or scripts inside a friend's object having full permissions granted

[12:17] Lomoco Binder: But that one doesn't mention deeding anywhere.

[12:17] Lomoco Binder: It might be the same fundamental issue though.

[12:17] Andrew Linden: ok well Lomoco, you can describe how to reproduce the bug, right?

[12:18] Lomoco Binder: Yes

[12:18] Simon Linden: I've seen that before - it's really annoying

[12:18] Andrew Linden: It is as simple as: make object with fully permissible notecards and then deed to group?

[12:18] Lomoco Binder: Yes

[12:18] Lomoco Binder: They then become unreadable while in the object's contents

[12:19] Andrew Linden: Hrm... yeah sounds like a bug.

[12:19] Andrew Linden: Notecards don't make much sense if they can't be read by anyone.j

[12:19] Lomoco Binder: I think you can drag them out of the objects contents and then access them fine

[12:19] Simon Linden: The jira makes an interesting point, however - are you expecting to be able to also edit that notecard?

[12:19] Lomoco Binder: But for sure not while they're in the object itself.

[12:20] Lomoco Binder: Ideally, you should be able to edit it, if such a group "role" exists,

[12:20] Lomoco Binder: Perhaps implement some new group roles to deal with accessing the contents of deeded objects?

[12:20] Motor Loon: is it a big hassel to just drag it to inventory, edit it and drag it back?

[12:21] Lomoco Binder: If you're doing lots of incremental changes, it can be.

[12:21] Qie Niangao: ah... the object itself has to have mod permissions to the next owner (that is, the group to which it's deeded) in order for it to be possible to set the embedded assets to share

[12:21] Andrew Linden: Ok, well I'll try to look into it. If you want to be able to track progress on it you'll have to make a jira.

[12:21] Lomoco Binder: Guess I'll have to go through support to try to get back on the Jira then, but thanks.

[12:21] Andrew Linden: Otherwise, if it ever gets fixed it will be done stealthily -- just show up someday.

[12:22] Lomoco Binder: I'll do some more testing and report on the jira.

[12:23] Qie Niangao: Unrelated begging: With normalmaps and spectralmaps coming, can we make sure there will be a script interface to setting, animating, aligning, scaling those guys?

[12:23] Andrew Linden: Huh, I wonder if they are planning on adding script interfaces for those.

[12:24] Andrew Linden: I don't know whether that will be done Qie. I'll have to ask.

[12:24] Qie Niangao: (I'd also like to beg for a script interface to projected texture lighting, but there are already two 2-year-old jiras asking for that)

[12:24] Qie Niangao: Thanks, Andrew.

[12:25] Motor Loon: agree on the projected lights thing

[12:25] Andrew Linden: In practice, it is often easier/faster for us server devs to add LSL calls than it is for us to get a working viewer with the UI we want.

[12:25] Andrew Linden: So often the LSL calls are done just so we can test the prototype.

[12:26] Andrew Linden: However, testing the materials stuff will already require a working viewer, so dunno if the same pressure will be there.

[12:27] Andrew Linden: Projected lights already exist right?

[12:27] Qie Niangao: yes, since 2010

[12:27] Motor Loon: err.. yeah

[12:27] Andrew Linden: So the ask is for LSL interface for that existing feature.

[12:27] Andrew Linden: Hrm... what parameters are set for projected lights at the moment?

[12:27] Simon Linden: What are the old jiras asking for it?

[12:28] Qie Niangao: hunting

[12:28] Qie Niangao: SVC-6390 was mine...

[12:28] JIRA-helper:

[#SVC-6390] Missing script interface to projected textures

[12:28] Draconis Neurocam: along with the normal light settings projected lights, have a texture, an fov value, a focus value, and an ambiance value

[12:28] Qie Niangao: scr-163 has more details.

[12:28] JIRA-helper:

[#SCR-163] New PrimitiveParams flag: PRIM_PROJECTOR

[12:29] NeoBokrug Elytis: New flags would be nice.

[12:29] Motor Loon: basically a llSetPrimitiveParams etc.

[12:29] Andrew Linden: What do you think Kelly?

[12:29] Simon Linden: I'm double-booked for another meeting I must attend -- take care everyone, see ya around

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

[12:30] Qie Niangao: thanks Simon

[12:30] NeoBokrug Elytis: Any updates on this one?

[12:30] JIRA-helper: [#SVC-6546] HTTP-IN URL's dropped after certain period

[12:30] Motor Loon: tc simon

[12:30] NeoBokrug Elytis: See yah.

[12:30] Draconis Neurocam: fov goes from 0.0, to 3.0, focus goes from -20.0 to 20.0, and ambiance goes from 0.0 to 1.0

[12:31] Kelly Linden: nope Neo.

[12:32] Kelly Linden: That one is very hard to repro.

[12:32] Motor Loon: ..and if you're messing with the primitiveparams anyway, I'd sure like a CURRENT_TEXTURE flag for all calls with a texture string (that just uses whatever is already on the primface)

[12:32] NeoBokrug Elytis: I understand. It doesn't happen often.

[12:33] Kelly Linden: Andrew: I can't think of a solid reason to not have the PRIM_PROJECTOR options. My only thoughts are on making sure there isn't a serious performance impact on the sim or viewer by changing those parameters rapidly. May want to get a graphics dev's opinion.

[12:34] Andrew Linden: Ok, I'll ask Runitai to see what he thinks.

[12:36] Motor Loon: its all viewer side anyway aint it?

[12:36] Andrew Linden: Yeah, the server relays the projector settings to the viewer, but doesn't use them otherwise.

[12:36] Kelly Linden: Motor: It may interact with interest list and amount of data that needs to be sent to viewers.

[12:36] Motor Loon: no different than change on regular lights via scripting

[12:37] Andrew Linden: Right, it probably wouldn't be different than a texture change on a face, for the server.

[12:37] Kelly Linden: may need some extra throttles to be sure no one floods a connection with it

[12:37] Motor Loon: sure

[12:37] Motor Loon: wont hurt

[12:37] Kelly Linden: it depends on what the existing implementation is.

[12:37] Kelly Linden: but yeah.

[12:38] Andrew Linden: ok I'll ask around for feasibility opinions

[12:38] Qie Niangao: I certainly have no problem with it being aggressively throttled, for any application I'd have.

[12:40] Qie Niangao: (same with the spectralmap / normalmap things... for which I'd be happy to make a jira, but won't if it's coming as part of the dev anyway.)

[12:41] Andrew Linden: Anybody have another topic for conversation?

[12:42] Andrew Linden: Alright, then perhaps we should end this user group early today.

[12:43] Motor Loon: I dont know if there's a jira for it... but I've been seeing alot of issues with (mabye interestlist related) stuff not rezzing fully until you interact with it (like rightclicking on it) and stuff not derezzing after leaving drawdistance... just hanging there... also until you interact with it

[12:43] Andrew Linden: Really Motor?

[12:44] Andrew Linden: There are two things that would make the object disappear if it is far enough away...

[12:44] Qie Niangao: I've seen that lately, too, the "click to see" effect... it's *instantaneous* when you click on it, though... doesn't feel like it's downloading it.

[12:44] Motor Loon: espicially that last bit is annoying.. I often sit online all day in a sim, and when I at some point look around I can see vehicles "hanging" all around me on the nearby sims... sometimes I've counted upto 20 objects that was there... but really wasnt (rightclick on them and they dissapear)

[12:44] Andrew Linden: the viewer will not render objects beyond its draw distance (I believe)

[12:44] Theresa Tennyson: I think there's a JIRA that says basically the HTTP texture system often doesn't realize it hasn't loaded the texture completely.

[12:45] Andrew Linden: and the simulator usually sends a KillObject message to things that it is no longer going to update should they move (removing from the server's "interestlist")

[12:45] Motor Loon: It's nothing like the "normal" behavior of stuff not rezzing or derezzing at distance...

[12:46] Andrew Linden: Ah Motor, when you click on them and they disappear... that sounds like you're suffering packet loss and losing some of the KillObject messages.

[12:46] Motor Loon: and its not textures in my case Theresa... its entire objects not derezzing as they should.... and for the rezzing part it's usually parts of a mesh object that dont rez until I rightclick it (atleast that where I've noticed it)

[12:46] Andrew Linden: Does the problem correlate with packet loss?

[12:46] Andrew Linden: And which viewer are you using when you see it happen Motor?

[12:46] Motor Loon: I normally dont have any packet loss

[12:47] Motor Loon: I'm mostly on Firestorm, but I see on on the LL beta also (those are the ones I use)

[12:47] Ashiri: Are the vehicles at region borders?

[12:47] Motor Loon: yes

[12:47] Qie Niangao: oh, *that* is a very old behavior

[12:47] Andrew Linden: There are a few operations that can cause bursts of packet loss...

[12:47] Motor Loon: very often the vehicles are hanging at simcrossings

[12:48] Andrew Linden: in particular, operations that might cause a big burst of UDP packets from the server

[12:48] Ashiri: like a vehicle crossing?

[12:48] Andrew Linden: Baker and I noticed this possibility when looking into the "cannot manage large groups" bug.

[12:48] Motor Loon: I should say this is not something that happens now and then... it's a daily thing for me.

[12:48] Theresa Tennyson: *cough*SVC-8124*cough*

[12:48] JIRA-helper:

[#SVC-8124] Excessive "ParcelOverlay reliable" messages sent by regions since last rolling restart (2012-08-08)

[12:49] Andrew Linden: Right Theresa, that ParcelOverlay bug does not help.

[12:49] Theresa Tennyson: That can cause HUGE packet loss.

[12:49] Motor Loon: Pretty sure I had it before that ParcelOverlay bug started too

[12:49] Andrew Linden: I'm just sayin' that while you might not get steady packet loss, you might get bursty packet loss.

[12:49] Andrew Linden: I've seen some content that can really burst too.

[12:49] Ashiri: I see 'hung' vehicle quite often in some regions

[12:50] Qie Niangao: right, the "vehicle crossing a sim border while out of view and getting stuck until clicked" problem has been around a long time... much longer than the "click to make object visible" thing that's happening the past few weeks.

[12:50] Andrew Linden: The last one I noticed was a dragon avatar that someone was wearing... it would burst full updates when crossing region boundaries.

[12:50] NeoBokrug Elytis: Happens with Avatars too, but to a lesser degree.

[12:50] Qie Niangao: Neo: yup

[12:51] Motor Loon: well, in any case - must be a solution for it?

[12:51] Andrew Linden: oh right... of course there would be a big burst of packets when a high-prim-count object crosses the region boundary

[12:51] Andrew Linden: the viewer must start getting updates from the new region

[12:52] Andrew Linden: as far as that region is concerned, this is a new object in that person's view

[12:52] Ashiri: ooh, blingtards stuck on region crossings!

[12:52] Andrew Linden: but some magic is done to let the viewer know not to delete its representation of the object as it crosses

[12:53] Andrew Linden: so the viewer can recycle some of its data and just hook up to the updates from the other region

[12:53] Motor Loon: so faulty magic mabye

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

[12:56] Motor Loon: here's what I'd see for example

[12:56] Andrew Linden: right, well... the bursty UDP packets could cause some KillObject messages to get dropped at that moment.

[12:56] Motor Loon points to textures infront of him

[12:57] Motor Loon: 2nd picture shows the effect of dragging a selection box around everything in the first picture

[12:57] Andrew Linden: the region border is in the middle of the shot?

[12:57] Motor Loon: yes

[12:57] Andrew Linden: yeah, well whenever the viewer asks to select an object that doesn't exist, we assume that the viewer doesn't know that the object is gone

[12:58] Andrew Linden: so we send a KillObject message in response.

[12:58] Andrew Linden: Hence it disappears on selection.

[12:58] Motor Loon: indeed

[12:58] Ashiri: ah!

[12:58] Motor Loon: would just be nice to get those automatically a little more stabile

[12:58] Youri Ashton: thanks for having us again Lindens!

[12:58] Youri Ashton: im off!

[12:58] Youri Ashton: bye everyone!

[12:59] Andrew Linden: right. I was working on the interestlist for a while... but got sidetracked by pathfinding maintenance

[12:59] Theresa Tennyson: Bye!

[12:59] Motor Loon: Should I try lower my bandwidth to see if that has an effect on it?

[12:59] Andrew Linden: hopefully I can return to that project soon.

[12:59] Motor Loon: I'm on a fiberoptic 50mbit connection so I usually keep it high, like at 1500kbps

[12:59] Andrew Linden: Sure Motor, I would guess that might actually reduce the problem, if the UDP system properly reduces its bandwidth.

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

[13:00] Motor Loon: yeah

[13:00] Motor Loon: Ill try and tell next time if I had an effect

[13:01] Andrew Linden: However, this duscussion does give me an idea to look into...

[13:01] Motor Loon: even better

[13:01] Andrew Linden: some UDP messages are sent totally "unreliable", which means that we don't actually care if they are lost or not.

[13:01] Andrew Linden: Others we do some tracking to try to figure out if they were lost.

[13:02] Andrew Linden: I wonder what method we're using to send the KillObject messages.

[13:02] Motor Loon: Think I can guess the answer to that one

[13:02] Andrew Linden: I'll try to look into that. It relates to the work I was doing on the interestlist anyway.

[13:02] Andrew Linden: ok thanks for coming everyone

[13:02] Motor Loon: thanks for having us

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

[13:02] Ashiri: Thanks !

[13:02] Yuzuru Jewell: Thank you!


Prev 2012.08.17 Next 2012.08.24