Simulator User Group/Transcripts/2012.06.22

From Second Life Wiki
Jump to navigation Jump to search


Prev 2012.06.19 Next 2012.06.26

List of Speakers

Andrew Linden Ardy Lay
Baker Linden GEE McAuley
Jonathan Yap Kelly Linden
MartinRJ Fayray Moundsa Mayo
Nalates Urriah Rex Cronon
Simon Linden TankMaster Finesmith
Widely Linden


[16:07] Andrew Linden: I was busy and forgot

[16:08] Simon Linden: We've got our fingers crossed on the OS upgrades -- they can have little surprises sometimes. Our initial channels, however, are looking good

[16:08] Ardy Lay: Tell the curious what OS you are installing?

[16:09] Simon Linden: On Wednesday I think the pathfinding code is going out, and the other channel will have the baked texture work

[16:10] Simon Linden: That's all I have ... is there anything else from other LIndens?

[16:11] Andrew Linden: Sigh... looks like I introduced some bug in pathfinding that Maestro just found. I've got to go fix it now...

[16:11] TankMaster Finesmith: so much for getting off work early

[16:12] Simon Linden: ok, good luck, Andrew

[16:13] Kelly Linden: I've been making some minor fixes to the estate access controls

[16:13] Kelly Linden: They might be interesting to discuss.

[16:13] Widely Linden: I know I had fun talking over them with you!

[16:13] Jonathan Yap: Are you working on the viewer UI too?

[16:14] Kelly Linden: One change is that currently the Access and Group Access (jonathan yea) lists are only effective if the region is not public access.

[16:14] Kelly Linden: I'm working on a tweak so that the Group and Agent access lists can allow you to override the options under 'Allow access only to Residents who'

[16:15] Jonathan Yap: Kelly, how does that change names in the Allow field whose names are in there so they can post event notices?

[16:16] Kelly Linden: So, for example, if you set an estate to 'Allow access only to Residents who: Have been age-verified', then those who you put in the Access list will be able to access the region even if they have not been age-verified.

[16:16] Kelly Linden: Jonathan: Does it make sense that anyone who could not access the region could create event notices?

[16:17] Jonathan Yap: It seems a bit odd, but that is also what names in Allow have been able to do

[16:17] Jonathan Yap: I can think of a use case--someone is an event manager, helps out with posting listings

[16:18] Kelly Linden: That tweak is server side, but the other tweak to go along with it will be to allow the Allowed Residents and Allowed Groups lists to be manipulated event if they are not currently "active"

[16:18] Kelly Linden: That is a fix for VWR-28068

[16:18] JIRA-helper:

[#VWR-28068] Estate access lists for groups and residents are blank when "Allow public access" is checked.

[16:19] Jonathan Yap: If you are poking around in that area in the viewer there is a jira to sort the names

[16:19] Kelly Linden: What jira?

[16:19] Jonathan Yap: I will try to find it, keep going while I look :)

[16:19] Rex Cronon: i am not familar with access lists, but can u search for a specific user name?

[16:19] Jonathan Yap: Rex, no, that is why having them sorted is helpful

[16:20] Simon Linden: No, there's no search into the lists

[16:20] Rex Cronon: then sorting would be quite usefull

[16:20] Jonathan Yap: vwr-29553

[16:21] Jonathan Yap: no jira helper?

[16:21] MartinRJ Fayray: VWR-29553

[16:21] Jonathan Yap:

[16:21] JIRA-helper: [#VWR-25993] Resident names not sorted in About Land floater's allowed/banned boxes

[16:21] Widely Linden: thanks Jonathan

[16:21] Jonathan Yap: I suspect the same problem applies to the estate names

[16:22] Simon Linden: I've been doing a little work in the "Top Scripts" window -- I'm working on adding the parcel name and script memory usage as two new columns

[16:22] Kelly Linden: I think that jira is for parcels - I have been working on the estate level controls Jonathan. I don't know how much overlap there is.

[16:22] Jonathan Yap: Kelly, if you think of it check to see if the estate names are not sorted too

[16:23] Jonathan Yap: I can't see that list so cannot tell

[16:23] Kelly Linden: :)

[16:23] Simon Linden: I'll bet under the hood they are the same type of UI widget

[16:23] Kelly Linden: yeah. I think it is whether or not we call 'sort()' after the data has arrived. :)

[16:23] Simon Linden: Right

[16:24] Ardy Lay: Is there some indication give when the data has all arrived?

[16:24] Kelly Linden: I'll have a look

[16:24] Ardy Lay: Last time I looked it was hard to tell when the list is complete.

[16:24] Jonathan Yap: They were sorted in V1

[16:25] Rex Cronon: did sim crash?

[16:25] Jonathan Yap: no, still here

[16:25] Ardy Lay: I think V1 sorted every time new data arrived.

[16:25] Kelly Linden: I need to head out early today. See you guys next week.

[16:25] Simon Linden: No Rex

[16:25] Jonathan Yap: bye Kelly

[16:25] Moundsa Mayo: Thanks, Kelly!

[16:25] Nalates Urriah: Thx Kelly

[16:25] Widely Linden: see ya kelly

[16:25] Rex Cronon: tc kelly

[16:25] MartinRJ Fayray: goodbye

[16:25] GEE McAuley: TC Kelly

[16:25] Simon Linden: The table is open for topics then

[16:26] Simon Linden: ... if we had a table

[16:26] MartinRJ Fayray: I found a way to add more than 500 people to the ban list...

[16:26] MartinRJ Fayray: which can lead to the entire list getting dropped

[16:26] Simon Linden: oh? How do you do that?

[16:26] Rex Cronon: u can put a huge tree trunk to serve as table

[16:26] MartinRJ Fayray: one second, I have a Jira

[16:27] Simon Linden: This sounds vaguely familiar ... we may have triaged it recently

[16:27] MartinRJ Fayray: I mean more than 300*

[16:27] MartinRJ Fayray:

[16:27] JIRA-helper: [#VWR-28780] It's possible to add more than 300 users to the parcel ban list.

[16:27] MartinRJ Fayray: sorry.

[16:28] MartinRJ Fayray: it just shot through my mind again:)

[16:29] Simon Linden reads ...

[16:29] MartinRJ Fayray: in some cases a full list can be forced to refresh, and while it is loading, you can add additional residents to the list.

[16:29] Simon Linden: Interesting ... I'm not too concerned about going a few over 300, but if that can cause any big drop of data from the list, that's a lot worse

[16:29] MartinRJ Fayray: I got to 305. In many tests the list dropped about 150 people from the list

[16:30] Simon Linden: Do you know if they were really dropped, or if the viewer might have just had incomplete data? Was it still small after re-logging?

[16:30] Meeter: Timecheck : User Group is half over

[16:30] MartinRJ Fayray: it was smaller after relogging, yes

[16:30] Simon Linden: ouch, that's not good

[16:31] MartinRJ Fayray: I had to add 150 again to test the case again

[16:31] MartinRJ Fayray: I tested that after some people were reporting emptied lists

[16:31] Jonathan Yap: I hope the database is not becoming corrupted with missing links or something

[16:32] Ardy Lay: There are almost thirty million accounts to choose from now so making a list of three hundred shouldn't be hard. ;-)

[16:32] Simon Linden: I think the database here is pretty simple, and the 300 limit is to keep the list size from getting too big

[16:33] Jonathan Yap: Is there something on the server to prevent > 300 entries from being added? Some non-conforming viewer could be programmed to exceed this limit.

[16:34] Simon Linden: But if it sent an update to the simulator that dropped half the list, that would definitely be something to fix

[16:34] Rex Cronon: u would have to enter 150 name again

[16:34] Simon Linden: I'd have to check, Jonathan, but we definitely try to enforce all limits server-side under the assumption that someone would change viewer limtis

[16:35] Ardy Lay: Uh, viewer doesn't send lists to the service, does it? I would expect it to just send adds and deletes.

[16:35] Simon Linden: right, that's very likely ... I'd have to check the code to be sure

[16:36] Simon Linden: I hope it's not sending up a big batch for removal

[16:37] Baker Linden: I'm working on testing out the new deployment stuff Simon talked about earlier, and I have to look at the viewer code to fully fix SVC-4968. The fix is in the sim code, but the viewer is still having problems displaying large groups.

[16:37] JIRA-helper:

[#SVC-4968] Group won't load - too many members

[16:37] Rex Cronon: is it sending names or sssid?

[16:38] Baker Linden: Was that for simon or me?

[16:38] Rex Cronon: either

[16:38] Simon Linden: I"m pretty sure it send names, Rex. Otherwise it will cause a huge batch of name lookups

[16:38] Ardy Lay: Baker, yeah, even the healthiest of workstations get grumpy with 90,000 names in a group.

[16:38] Baker Linden: well, it's around 16k when it craps out

[16:39] Ardy Lay: Heh, okay. Have you noticed any packet loss as a cause or as a result?

[16:39] Rex Cronon: to me it would make sense to send uuid, as they r of fixed length

[16:39] Baker Linden: not from what I can see anymore. My sim-fix should prevent that

[16:39] Ardy Lay: Packets arriving "out of sequence" seem to busy up the viewer.

[16:39] Baker Linden: that shouldn't happen anymore

[16:39] Ardy Lay: Nice. Thanks

[16:39] Jonathan Yap: Wasn't one issue was that a certain part of the group had to load (role data) before the names would be processed?

[16:40] Baker Linden: I remember reading that in the Jira, and I had all 40k members sent in a group send to the sim without issue

[16:40] Baker Linden: yes, Jonathan

[16:41] Jonathan Yap: Is that data going by UDP? What happens if there is a packet dropped ?

[16:41] Jonathan Yap: anyhow, have fun Baker, a lot of people will appreciate your work

[16:41] Baker Linden: no, it's going through HTTP

[16:42] Baker Linden: The problem was that the names would take a long time to get and send back, and cause stalling. The group info on the viewer was waiting for a ton of stuff, and since we're choking the back-end, it can't send information. I batched it up a bit better so hopefully it won't choke the back-end anymore

[16:43] Baker Linden: I rambled a bit there, if that didn't make sense. It's been a long day. :)

[16:43] Jonathan Yap: we followed

[16:43] Simon Linden: Have any of you large group users noticed a change in group notice delivery or new problems sending chat? I had someone complaining the other day and wondered if it was just an individual case or something more widespread

[16:44] Rex Cronon: why not make a zip file with names for each group and just send it?

[16:44] Jonathan Yap: I saw a jira pass by about group notices being delivered twice I think

[16:44] TankMaster Finesmith: weve seen reports of some users not being able to send GN ot they werent getting delivered

[16:44] Valinye: I've had recent reports of notices not being received

[16:45] Nalates Urriah: I still have the open quickly or the attachments fail to open or save...

[16:45] Jonathan Yap: Nalates, that is a deep issue not easily fixed, and also partly a viewer UI problem

[16:45] Simon Linden: I've also seen that bug where a notecard will fail to open, giving a permissions warning but then opens on the 2nd try

[16:46] Simon Linden: I sounds like there might be something going on, but it's not a big increase in problems that is significantly worse than before

[16:46] Jonathan Yap: Nalates, what is happening is the transaction ID is becoming stale (or maybe more than that)

[16:46] Baker Linden: Rex -- Andrew and I thought of that as well -- if this doesn't work well, we might try something like that

[16:47] TankMaster Finesmith: VWR-29148

[16:47] Simon Linden kicks the jira bot

[16:47] Rex Cronon: good luck:)

[16:47] Jonathan Yap:

[16:47] JIRA-helper: [#STORM-1890] Enabling "Only friends and groups can call or IM me" mutes your own chat in all groups and mutes all group chat from non friends

[16:47] TankMaster Finesmith: thats the same one

[16:47] Jonathan Yap: that got moved and fixed

[16:47] Baker Linden: There's probably much more going on that what I've done, but my changes seems to have sped things up sim-side a bit

[16:47] TankMaster Finesmith: ok

[16:48] TankMaster Finesmith: says unresolved though

[16:48] Jonathan Yap: martin is on it

[16:48] MartinRJ Fayray: the fix is on bitbucket

[16:48] Jonathan Yap: I guess he did not push start progress (and he may still be banned from jira, so cannot push the button)

[16:48] MartinRJ Fayray: I have to push start progress?

[16:49] Jonathan Yap: yup!

[16:49] MartinRJ Fayray: ok. done :D

[16:49] Jonathan Yap: now transition it to the next stage

[16:49] Simon Linden: Do you know if that's in the viewer pipeline now?

[16:49] MartinRJ Fayray: I don't think so. but Oz Linden knows that it's fixed

[16:50] Jonathan Yap: It's not going out in the upcoming viewer release

[16:51] Jonathan Yap: martin, if you have finished that now push ready for review and fill in the bitbucket and code review links

[16:51] MartinRJ Fayray: ok I'm on it. ty Jonathan!

[16:51] Jonathan Yap: the code review field is now mandatory

[16:51] MartinRJ Fayray: done

[16:52] GEE McAuley: Thanks Martin for looking into that.

[16:52] Jonathan Yap: Tank, I think you missed Oz's last meeting where this was brought up

[16:52] Simon Linden: cool, I'm going to start sending viewer bugs to you guys

[16:53] Jonathan Yap: Send them to Oz and he knows who to parcel them out to

[16:53] Simon Linden: right, I was joking

[16:53] MartinRJ Fayray: hehe

[16:53] Jonathan Yap: haha, I wasn't :)

[16:53] MartinRJ Fayray: more than welcome, Gee

[16:53] TankMaster Finesmith: yeah, i wasnt home yesterday

[16:53] TankMaster Finesmith: so i missed Oz's meeting

[16:54] Ardy Lay: still hiding in the shadows

[16:54] JIRA-helper: [#VWR-23393] Viewer reports script info for the wrong parcel, in some cases

[16:54] TankMaster Finesmith removes all scripts

[16:54] TankMaster Finesmith: there, issue resolved

[16:55] Simon Linden: Is that easily reproducible, or does it happen randomly?

[16:55] Jonathan Yap: hmmz, what happens if you have an object who has a scripted prim extending into a no script parcel?

[16:55] Meeter: Timecheck : User Group is almost over

[16:56] Ardy Lay: It seems to be dependent on parcel shape - fragmentation

[16:56] Simon Linden: I believe all the script usage and such are based on the root prim location -- or should be, at least

[16:56] Rex Cronon: or a scripted object that has multiple prims on different parcels?

[16:56] Ardy Lay: There are some linked issues, one is in SEC

[16:58] Ardy Lay: SCR-133, MAINT-357 Oh, not in SEC, in SCR

[16:58] JIRA-helper:

[#SCR-133] Script Info reporting on discontiguous parcels is not working as expected.

[16:58] Rex Cronon: i wonder if it would be possible once u guys fix a SEC jira, that u can make it public? i mean is already fixed. right?

[16:58] Ardy Lay: Some of them still have unpublishable data in them.

[16:58] Rex Cronon: ah. ok

[16:59] Jonathan Yap: But there might be similar exploits, or OSGrid suffers from the same problem, better to keep it private

[16:59] Simon Linden: I can't think of a good reason to make it public, and a few not to

[16:59] Rex Cronon: np. just asking

[17:00] Ardy Lay: I was thinking about the collateral data that gets into those. They were not written for publication so sensitive information ends up in them.

[17:00] Jonathan Yap: Thank you Simon and Andrew (if you are here)

[17:00] Meeter: Thank you for coming to the Server User Group

[17:00] Simon Linden: Right, they often have comments that were made expecting it to stay private

[17:01] Simon Linden: We're out of time ... thank you all for coming and the good chat

[17:01] TankMaster Finesmith: have a great weekend, everyone

[17:01] Moundsa Mayo: Andrew, Simon, Baker, Widely, thanks for your time, the information you share, and all your hard work!

[17:01] Ardy Lay: Thank you

[17:01] Valinye: THank you Simon

[17:01] Nalates Urriah: Thx Simon

[17:01] MartinRJ Fayray: goodbye everyone, have a good weekend

[17:01] Jonathan Yap: oups, thank you as well, Widely!

[17:01] GEE McAuley: Take Care all Have a great weekend.

[17:02] Jonathan Yap: and Baker, geesh, my mind is going

[17:02] Rex Cronon: tc Andrew, Simon, Baker, Widely, and everybody here. and have u good day:)

[17:02] Jonathan Yap: too hot here

[17:02] Moundsa Mayo: Up.Up, and ... oops

[17:03] Simon Linden: bye everyone -- see you next time

[17:03] Widely Linden: thanks guys!


Prev 2012.06.19 Next 2012.06.26