Simulator User Group/Transcripts/2014.02.25
|Prev 2014.02.18||Next 2014.03.04|
List of Speakers
|Baker Linden||Inara Pey||Jenna Felton|
|Jonathan Yap||Kallista Destiny||Kelly Linden|
|Kennylex Luckless||Lexbot Sinister||Lucia Nightfire|
|Mona Eberhardt||Nalates Urriah||Oz Linden|
|Qie Niangao||Rex Cronon||Simon Linden|
|Skills Hak||Tankmaster Finesmith||Whirly Fizzle|
[12:01] Simon Linden: Let's get going with the news ... the server main channel was rolled out this morning
[12:02] Simon Linden: It was a little slower than usual, and started earlier, as we've been adjusting some of the ways that's done. It will get faster again when we sort out a few things
[12:02] rcdsQueChatLog: Rex Cronon, you can read the log here:
[12:03] Simon Linden: but if you're really paying attention you'll notice regions didn't move from the server they were shut down on
[12:03] Jonathan Yap: Hippotropolis took over 30 minutes to restart; is that part of the changes?
[12:03] Nalates Urriah: ??? You're change so they restart on the same server?
[12:03] Simon Linden: Tomorrow morning the RC channels will get updates ... one of these _should_ be invisible, as it's just some system update
[12:04] Simon Linden: That's because of the way it was rolled this morning, Jonathan. We don't want to do that in the future
[12:05] Mona Eberhardt: Simon, last week, this was mentioned by Stickman Ingmann: "Stickman Ingmann: Speaking of llSetPrimitiveParamsFast. I've been noticing that objects made of numerous prims haven't been changing all of the prims lately. Even when done in a single call. How is such a change sent to the client? Is there a message size cap? Is it sent as a bunch of individual packets, one per prim, and some are getting lost? It's random for everyone looking at the object how many and which prims don't change." - Has anything been done to counter it?
[12:05] Simon Linden: The code going into Magnum has some of the Sunshine code that will kick in with their project viewer ... it's using the new inventory API which will hopefully help prevent clothing malfuctions
[12:06] Simon Linden: ... and make inventory operations more efficient
[12:06] Simon Linden: Mona - we haven't worked on that this past week
[12:07] Simon Linden: I'm not sure when it might get attention
[12:07] Tankmaster Finesmith: for those that compile FS, last night I updated our -exp repo which is that same sunshine code base to the latest LL has to offer for it
[12:07] Inara Pey: That code on Magnum is being enabled, rather than deployed, right? Nyx and Oz were under the impression the code had been previously deployed disabled.
[12:07] Tankmaster Finesmith: our next release wont contain the AIS3 stuff, however
[12:08] Tankmaster Finesmith: correct, inara
[12:09] Simon Linden: Yes, Inara .. it's really just turning on a switch, but it's effectively the same thing. That code didn't run previously
[12:09] Simon Linden: If things blow up, we can turn it off without a full re-deploy
[12:09] Inara Pey: Yup. Just that the last time it came up elsewhere, there was a tad bit of confusion ?
[12:10] Tankmaster Finesmith: it was never mentioned anywhere that the code had been deployed, that was the confusion
[12:10] Simon Linden: yes ... it's definitely confusing if we talk about code being deployed that's actually turned off ... it might be there but it's not workin
[12:10] Simon Linden: g
[12:11] Inara Pey: That was my point, Tank....
[12:12] Kallista Destiny: on the mchine but not running is not really deployed.
[12:12] Mona Eberhardt: Hello Qie.
[12:13] Qie Niangao: Hi Mona, all.
[12:13] Inara Pey: Hey, Qie!
[12:13] Rex Cronon: hi qie
[12:14] Simon Linden: Mona - for some of the details about your question, some updates such as the terse ones are packed together in the same packet. So if one packet gets lost, it may contain updates for more than one prim
[12:14] Mona Eberhardt: I see.
[12:14] Mona Eberhardt: It only started happening last week.
[12:15] Simon Linden: I don't remember anything changing in the update code recently
[12:15] Whirly Fizzle: Huh? Are you sure? Its an oldish issue from about feb last year
[12:15] Simon Linden: The http viewer code just went out ... I wonder if a different http traffic profile would change how it conflicts with the udp traffic
[12:15] Rex Cronon: shouldn't the viewer know when it is missing a packet?
[12:16] Simon Linden: yeah, but stuff happens :P
[12:16] Tankmaster Finesmith: not if its udp
[12:16] Tankmaster Finesmith: thats the main issue with udp, if a packet is lost, theres no record of it
[12:16] Rex Cronon: there r ways to get around that
[12:17] Lucia Nightfire: overhead
[12:17] Simon Linden: There are packet numbers in the headers we compose, so there's some knowledge that something got lost
[12:17] Tankmaster Finesmith: ll has added code to mitigate a lot of that, but its not perfect
[12:17] Kallista Destiny: But that is just recreating the process of TCP/IP
[12:17] Simon Linden: for some things the viewer / simulator will re-try sending packets
[12:17] Mona Eberhardt: Well, it started affecting me last week, Whirly. There was, however, a behaviour that's similar to it, which I've noticed back in 2012: When I have a seriously multi-prim (some 100+ prims) object and try to make all of the linkset full bright, not all of the prims will get the attribute.
[12:17] Simon Linden shrugs ... it's what SL does, dig into the viewer code if you want the real details.
[12:18] Kallista Destiny: Yeah sometimes it's not worth sending a secod old position message
[12:18] Simon Linden: Right Mona, that's the symptom
[12:18] Mona Eberhardt: Hi Oz.
[12:18] Tankmaster Finesmith: right, hall
[12:18] Whirly Fizzle: Heya oz
[12:18] Mona Eberhardt: But this time, Simon, even linksets with few prims in them are affected.
[12:18] Kallista Destiny: Ave Oz
[12:18] Tankmaster Finesmith: for some things, ud is grate, cus if the packet is lost, its out of date by the time it would be r3sent
[12:18] Oz Linden: hi all
[12:18] Tankmaster Finesmith: udp*
[12:18] Rex Cronon: hi oz
[12:18] Tankmaster Finesmith: hi Oz
[12:19] Mahadma (mahadma): Greetings :)
[12:19] Rex Cronon: dir u bring dorothy too;)
[12:19] Tankmaster Finesmith: oh, oz, i got HTTP into FS last night
[12:19] Simon Linden: To really isolate that, it would be nice to have a case where it's reproduced, then actually record what gets updates and what doesn't ... that would be a big step to confirming the theory
[12:19] Baker Linden: holy crap it's Oz!
[12:19] Tankmaster Finesmith: just fyi
[12:19] Oz Linden: nice
[12:19] Simon Linden: Welcome, Oz :)
[12:20] Tankmaster Finesmith: hi skilz
[12:20] Skills Hak: hellos
[12:20] Kallista Destiny: Now all we need is for the 64bit build to complete (it failed last night)
[12:20] Rex Cronon: wow. baker is here too:)
[12:20] Skills Hak: the whole gang
[12:20] Rex Cronon: hi skills
[12:22] Lexbot Sinister: and suddenly, crickets ;)
[12:23] Lexbot Sinister: XD
[12:23] Rex Cronon: lol. i thought i was diconneted. was ready to relog
[12:23] Seewo (siwo): :P
[12:23] Qie Niangao: Oh, what the heck, as long as there are Lindens aplenty afoot: Any news about how / when the jira will be re-opened?
[12:23] Kennylex Luckless: BAKER!
[12:23] Baker Linden: Well, I'm filing for my server-side code to go out to aditi -- I'll have ti filed for release by end of week :)
[12:23] Tankmaster Finesmith: group banz!?!?
[12:23] Baker Linden nods
[12:23] Whirly Fizzle: \o/
[12:24] Baker Linden: We're testing it for compatibility with old viewers and such
[12:24] Kennylex Luckless: Now we just neew automatic griefer ban.
[12:24] Tankmaster Finesmith: if it brakes phoenix, ship it out as fast as you can :P
[12:24] Baker Linden: and then once that's stable and I know I didn't break anything, I'll start filing for the viewer to become a release candidate
[12:24] Kallista Destiny: Or sooner even
[12:24] Whirly Fizzle: Like it could get any more broken? :P
[12:24] Tankmaster Finesmith: lol
[12:24] Baker Linden: fair enough
[12:24] Baker Linden: lol
[12:25] Rex Cronon: don't temp fate whirly:)
[12:25] Rex Cronon: tempt*
[12:26] Tankmaster Finesmith: i have a feeling its going to be a lengthy RC process with all the other stuff in the queue
[12:26] Rex Cronon: btw. when we get those notices that somebody paid us money, could it alsow include on what grid was the payment made?
[12:26] Baker Linden: yeah, there will be plenty of time to integrate the viewer code into FS
[12:27] Tankmaster Finesmith: crash stat update, interest list, AIS3 (soon) group bans (soon), viewer-bear...
[12:27] Baker Linden: group chat
[12:27] Tankmaster Finesmith: ?
[12:27] Baker Linden: is still broken
[12:27] Lexbot Sinister: are you gonna breakl group chat even more???
[12:27] Whirly Fizzle: Yes, still broken :P
[12:27] Baker Linden: is the end of that sentence
[12:27] Tankmaster Finesmith: lol
[12:28] Baker Linden: SIKE
[12:28] Lexbot Sinister: Well, very broken this weekend.
[12:28] Rex Cronon: for a sec i thought u were going to fix it baker;)
[12:28] Lexbot Sinister: I would like to add that there is no way to report it to support, either.
[12:28] Tankmaster Finesmith: i thought it was "is still broken" comes before" group chat"
[12:28] Tankmaster Finesmith: :D
[12:28] Baker Linden: haha
[12:29] Simon Linden: Rex ... go ahead and file a jira on that, but I don't think we'd want to have any grid info in the usual payment message. It might be good to have it say something for a place that's not the main grid, but otherwise it's likely to confuse the majority of people
[12:29] Whirly Fizzle: Yeah, whats going on with that? The last few weeks, pretty much most weekends a group chat server has been totally down. This week it affected groups starting with 3 in their key. This never used to happen as regularly.
[12:29] Baker Linden: weird
[12:29] Tankmaster Finesmith: i remember LL fixed group chat real good in phase one, then a week later when they implented phaze 2, it broke again
[12:30] Lexbot Sinister: Yeah, i asked in the jira where to ask for support (because it might be just something needing a reboot) but usual support options are not appliable
[12:30] Simon Linden: We would all love to fix group chat :)
[12:30] Tankmaster Finesmith: ho hard can it be? :P
[12:30] Lexbot Sinister: No but.. it was totally down. Not a little lag or more lag, just off. Entirely
[12:30] Simon Linden: Hard enough, lol
[12:30] Tankmaster Finesmith: hehe
[12:31] Whirly Fizzle: Yesh., totally dead as doornail for groups with key starting with 3
[12:31] Lexbot Sinister: right.
[12:31] Kallista Destiny: perhaps more that hard enough, NP complete?
[12:31] Mona Eberhardt: Maybe it can be fixed by a very angry Ritchie Blackmore... :/
[12:31] Rex Cronon: i don't think group chat is ever going to be fixed:( because its impossible for over 500 people to talk and understand eachother if they do it all at the same time
[12:31] Lexbot Sinister: We can ask support to restart a sim, but we can't ask for restart of chat server
[12:31] Simon Linden: it sounds like the system monitoring tools aren't picking up that case where one of the chat servers gets in trouble ... I wonder if the server host is still running fine but just the chat part is busted and not being detected
[12:31] Skills Hak: are the groups spread out over tables or whole different databases on different servers?
[12:31] Simon Linden: Yes Mone :)
[12:32] Skills Hak: haha
[12:32] Skills Hak: mm sql
[12:32] Tankmaster Finesmith: ll has a lot of servers to handle group chat
[12:32] Kallista Destiny: I can think of a fix, double the number of chat servers, and use one more bit to select them.
[12:33] Seewo (siwo): May I a have a question? ...is there something new about https://jira.secondlife.com/browse/SVC-422 ?
[12:33] Lucia Nightfire: public selling won't eb goign away any time soon
[12:34] Lucia Nightfire: I didn't even open that link but I can tell what it is
[12:34] Seewo (siwo): heh
[12:34] Simon Linden: There's no news on that, Seewo
[12:34] Skills Hak: lol 2007
[12:35] Whirly Fizzle: The group chat servers going down at the weekends started in Oct 2013. If thats any use for timing.... and by doen, I mean totally down for certain groups.
[12:35] Seewo (siwo): that's the sad thing about... 2007 and still nothing :/
[12:35] Whirly Fizzle: *down
[12:35] Rex Cronon: maybe ther r too many robots in them listening?
[12:36] Mahadma (mahadma): It is a good idea in my opinion
[12:36] Lexbot Sinister: hmmm
[12:37] Lexbot Sinister: i'd add that a prcel could be set to sell from either group/owner
[12:37] Lucia Nightfire: you mean public./group
[12:37] Simon Linden: Are you talking about a land sale, Lex?
[12:37] Lexbot Sinister: No
[12:37] Lucia Nightfire: they mean objects for sale
[12:37] Lucia Nightfire: in sandboxes
[12:37] Lexbot Sinister: just making it impossible for other people to sell stuff on my own parcel
[12:37] Simon Linden: ok, right
[12:38] Lucia Nightfire: or left on 0 autoreturn parcels, etc
[12:38] Lucia Nightfire: "public selling"
[12:38] Lexbot Sinister: would get rid of the prim over vendor scam
[12:38] Seewo (siwo): ^this
[12:38] Skills Hak: would it?
[12:38] Lucia Nightfire: not really, you still get that in other forms
[12:38] Skills Hak: you can just attach alpha prims to your avatar center and stand somewhere close
[12:39] Lexbot Sinister: so? if you can't sell or be paid on that parcel, you can't ;)
[12:39] Skills Hak: what about stores :o
[12:39] Tankmaster Finesmith: :o
[12:39] Lexbot Sinister: belong to land group.
[12:39] Skills Hak: it's kinda the same thing with misleading ads on web stores
[12:39] Skills Hak: people gotta open their eyes
[12:40] Lexbot Sinister: yes, but we should also minimise the most obvious ways of scamming, too
[12:40] Rex Cronon: i don't think opening your eye will help u see something that is invisible:)
[12:40] Simon Linden: It wouldn't be a perfect solution to all the scams out there, of course, but it would help against some
[12:41] Seewo (siwo): but... afaik attachments can't have buy/pay option :P Only pay to avatar, and that's rather different window
[12:42] Skills Hak: you could still rez your alpha fake vendors on another parcel nearby too
[12:42] Lucia Nightfire: unless you're referencing root prims, heh
[12:42] Rex Cronon: wait a sec. don't get a confirmation dialog each time u pay an object that asks u if u want to pay a specific amount to an object belonging to a specific user?
[12:42] Seewo (siwo): yes
[12:42] Lucia Nightfire: but there are other scams than paying objects with covering prims, heh
[12:42] Simon Linden: Yes Rex but it's easy to scam object and account names to look very similar to the target
[12:42] Lucia Nightfire: or rather scms, threats
[12:42] Seewo (siwo): but who pays attention to it...
[12:42] Lexbot Sinister: right. I was just thinking of the iterations of "can't be paid because on parcel2 and it didn't look so good in my head after all. But for rezzed objects? I wonder why it's not the norma already
[12:43] Lexbot Sinister: would be also interesting when wanting to enforce a "residental area only" policy
[12:43] Lexbot Sinister: and lower tier for it \o/
[12:44] Lucia Nightfire: Simon, Kelly, How easy would it be to implement this?: https://jira.secondlife.com/browse/BUG-4329
[12:44] Lucia Nightfire: prim hostless sensor capability, basically
[12:45] Rex Cronon: what is that for lucia?
[12:45] Lucia Nightfire: I have two gadgets a defnse, offense and a security app that could benefit from it
[12:46] Lucia Nightfire: presently you have to have a prim host to fire a sensor, it uses the host's pos and rot as ref, I'd like to be able to input into the sensor function itself w/o relying on an actual prim's presence/pos and rotation
[12:46] Kelly Linden: llSensor and llSensorRepeat are quite encumbered and difficult to modify or mimic
[12:47] Rex Cronon: i don't get it. what would an hostless prim do, or not do?
[12:47] Mona Eberhardt: I don't understand what a hostless prim is. :/
[12:47] Lucia Nightfire: I can fire a sensor a inputted pos and rot w/o pos/rot a prim there
[12:47] Skills Hak: a hostess prim is a prim who receives or entertains other linksets
[12:48] Simon Linden: lol
[12:48] Seewo (siwo): :D
[12:48] Lucia Nightfire: one app I use has multiple sensors taht repeatedly pos themself around a sim, I'd like to be able to do all taht with one script in a stationary prim
[12:48] Lucia Nightfire: another app I want to do it from an attachment
[12:49] Kennylex Luckless: Btw, is there anyway for Linden to stop objects that has so large area that they lag sim or crash gfx cards?
[12:49] Skills Hak: i have a funny trick to bypass sensors :o
[12:49] Lucia Nightfire: agent sensors yes, heh
[12:49] Skills Hak: but it's silly
[12:49] Rex Cronon: oh. so u want to scan at a specfic pos that can be anywhere in the sim and is not tied to the location of the prim that has the script in it?
[12:50] Qie Niangao: I'm assuming this sensor needs to be a sensor, can't be llCastRay.
[12:50] Lucia Nightfire: right qie, cause cr is actually limited in several apps
[12:50] Lucia Nightfire: esp agent hit boxes, lol
[12:50] Qie Niangao: indeed. but it does behave as "hostless" in your sense.
[12:51] Lucia Nightfire: I have one app that I use it when I alt cam someone, but if they are slightly moving or sitting on something its useless
[12:51] Tankmaster Finesmith: kenny, i thik the viewer has that capibility already
[12:51] Qie Niangao: yep
[12:51] Simon Linden: Kennylex - there are actually a debug limit in the viewer that try to avoid some of those graphics crashers. Check out RenderAutoHideSurfaceAreaLimit
[12:51] Tankmaster Finesmith: long enough name? :P
[12:51] Kennylex Luckless: I has never get them to work.
[12:52] Kennylex Luckless: and those objects should be removed ASP from servers.
[12:52] Simon Linden: It's not long compared to the length of that Debug Settings pop up menu :)
[12:53] Rex Cronon: lucy u need llSensor( vector scanPos, string name, key id, integer type, float range, float arc );
[12:53] Rex Cronon: that wouild be quite useful:)
[12:53] Rex Cronon: would*
[12:53] Lucia Nightfire: I outlined the preferred criteria in the jira rex
[12:53] Lucia Nightfire: if you can see it forgot it you can or not
[12:54] Lucia Nightfire: I could outline all teh apps in that jira, but some of them I consider a competitive ideas, heh
[12:54] Lucia Nightfire: in a way, revolutionary if I had that ability
[12:54] Lucia Nightfire: nothing out of this world, but neat nonetheless
[12:55] Rex Cronon: i can't see it
[12:55] Lucia Nightfire: Proposed function syntax:
llSensor2(string name, key id, key owner, integer type, float range, float arc, vector pos, rotation rot);
llSensorRepeat2(string name, key id, key owner, integer type, float range, float arc, float rate, vector pos, rotation rot);
[12:55] Lucia Nightfire: there ya go
[12:55] Rex Cronon: :)
[12:55] Kelly Linden: that is quite the list of args
[12:56] Simon Linden: it makes some sense .... it possibly opens up parts of a region where you currently couldn't reach with sensors, however, but that requires a special layout
[12:56] Lucia Nightfire: the pos and rot input are necessary at least, the additional extra was teh owner filter
[12:56] Lucia Nightfire: since we don't have one in the current functions
[12:56] Lucia Nightfire: it would facilitate llDetectedOwner() though
[12:56] Rex Cronon: i think u can implement that kelly:)
[12:57] Lucia Nightfire: anyway, figureed I'd ask, hehe
[12:57] Kelly Linden: Adding an owner filter would be very difficult, especially for llSensorRepeat. I'd have to do some digging but even the position / orientation args may also be prohibitively difficult in the llSensorRepeat case.
[12:58] Lucia Nightfire: idk how its currently plumbed, I'm sure theres a nesting to a degree, lol
[12:58] Kelly Linden: llSensorRepeat involves cross sim messaging which makes things complicated.
[12:59] Simon Linden: Limitiing it to one region makes it much, much easier
[12:59] Lucia Nightfire: yeah, in a way, I was never fond of returns taht were foudn out of the region or beyond the range I set
[13:00] Lucia Nightfire: I always disregard those in teh script
[13:00] Qie Niangao: Yeah, historically, I seem to recall some problems with accuracy of other-sim returns.
[13:00] Lucia Nightfire: not sure if anyoen would need them though, just my thoughts
[13:00] Kelly Linden: I've also been a bigger fan of non-sensor methods for finding objects or finding information about objects. llSensor has a lot of baggage that isn't actually desired in many cases.
[13:00] Jenna Felton: most things are limited to the region, the script is running + 10m inside connected regions, not farther
[13:01] Lucia Nightfire: https://jira.secondlife.com/browse/SCR-451 <--then you may like this request, heh
[13:01] Kelly Linden: Unfortunately I have a 1pm I need to get to. Thanks for coming everyone. :)
[13:01] Tankmaster Finesmith: tc kelly
[13:01] Tankmaster Finesmith: thx for coming!
[13:01] Yuzuru Jewell: Thank you, Lindens.
[13:01] Lucia Nightfire: later
[13:01] Rex Cronon: tc kelly
[13:01] Whirly Fizzle: Ty Lindens
[13:01] Nalates Urriah: Thx Lindens
[13:01] Tankmaster Finesmith: have a good one, simon
[13:01] Mona Eberhardt: Thank you Lindens.
[13:01] Kennylex Luckless: Next time drag Ebbe here
[13:02] Tankmaster Finesmith: and baker, congratz on the promotion with group banz !
[13:02] Simon Linden: I need to go too ... maybe for the sensor functions thinking about it with a completely blank slate would be a good idea ... like what would the ideal function be?
[13:02] Rex Cronon: tc simon
[13:02] Simon Linden: anyway, thats food for thought ... thank you all for coming this week
|Prev 2014.02.18||Next 2014.03.04|