Chat log from the meeting on 2017-06-20

From OpenSimulator

Jump to: navigation, search

[11:00] Arielle.Popstar summertime does affect attendance though as more people outside and holidays
[11:08] Arielle.Popstar i see everyones name ok but i am the only local resident
[11:09] Gavin.Hird yup. any reason for the slowness?
[11:09] Arielle.Popstar i blame mesh
[11:09] Leighton.Marjoram not sure what it is, it only happens once in a while and some times when I arrive.
[11:09] Leighton.Marjoram Its disorientating when I can see the names above peoples heads but not in local chat.
[11:10] Arielle.Popstar i have bubble chat on to avoid that problem
[11:10] Don.Smith i was at the other meeting place, i didnt know about this one
[11:10] Leighton.Marjoram That is my usual strategy @Arielle
[11:11] Don.Smith << i'm not in the cool kids club anymore, lol >
[11:11] Arielle.Popstar oh the sign is no longer up?
[11:11] Arielle.Popstar that says the dev meeting is here?
[11:11] Don.Smith I did not see one
[11:11] Arielle.Popstar ok
[11:11] JayR Cela: yes this location is very difficult to find :_(
[11:11] Kayaker Magic: Yeah, I looked last week for a link and didn't see one at the old site.
[11:12] George Equus: sign is gone
[11:12] George Equus: been for a few weeks
[11:12] JayR Cela: the OSGrid info page still just lists it's location as in Wright Plaza
[11:12] Selby.Evans Be sure to get LM for here if you don't have one.
[11:12] George Equus: and a rather cryptic name  :)
[11:12] Don.Smith ty, George gave me a lm =)
[11:13] Arielle.Popstar i am not sure how long this region is up for but was told it is better for meetings
[11:14] Arielle.Popstar some people had issues with wright because it has so much stuff it created lag
[11:14] Leighton.Marjoram this is a nice low impact region
[11:14] Selby.Evans Does seem to have less lag
[11:14] Don.Smith agreed
[11:14] George Equus: On request by Ubit I think, he suffer lag at the other place...
[11:14] Arielle.Popstar yes
[11:14] George Equus: personally never suffer much lag on wright
[11:14] Arielle.Popstar i used to on a low bandwidth connection
[11:15] Selby.Evans If you ever need a place, I can provide space on Kitely
[11:15] George Equus: Question... any recent (like today) problems with HG?
[11:15] JayR Cela: @George / not for me :_)
[11:16] Leighton.Marjoram not that I noticed George
[11:16] Marcus.Llewellyn Coming here was my only HG jump today. Went fine, once I fixed my URL typos. ;)
[11:16] Arielle.Popstar not been on HG for a week or so
[11:16] Don.Smith i tp'd from kitely to OS easily
[11:16] George Equus: got stuck everywhere all afternoon when trying to HG
[11:16] Selby.Evans I have not seen any HG problems recently.
[11:16] George Equus: OK, some local thing then...
[11:16] Don.Smith i did notice in the last few days, many of my HG lm's "cant contact the region"
[11:16] George Equus: US Euro connection
[11:17] Arielle.Popstar thats a pain Don
[11:17] George Equus: or can't find it at all
[11:17] Don.Smith or even show on map
[11:17] Selby.Evans Except with Outworlds -- it has dynamic dns
[11:17] George Equus: even got stuck trying to tp from one level to another on my own region
[11:17] Don.Smith ouch
[11:17] Marcus.Llewellyn Starting to seem like it is going to be a user meeting rather than a dev meeting today. ;)
[11:17] George Equus: mmmm unexpexted
[11:18] George Equus: sorry. please over to Devs
[11:18] Arielle.Popstar they can read it after and see what there are issues with etc
[11:18] Kayaker Magic: I thought it was an "open house" meeting with the devs
[11:21] Arielle.Popstar so any other issues or things that need fixing we can talk about?
[11:21] Leighton.Marjoram /puts thinking hat on
[11:21] Selby.Evans The OS Devs could have their own WG.
[11:22] Selby.Evans The IMA WG support is working now -- The OS Devs could have their own WG.
[11:22] Arielle.Popstar WG?
[11:22] Don.Smith I am hopeful Ubit can share a possible fix for a texturing issue i've had for a while
[11:22] Selby.Evans working group
[11:22] Leighton.Marjoram Working Group (WG) I assume
[11:22] Arielle.Popstar ahh
[11:22] Arielle.Popstar turns on lightbulb
[11:23] Arielle.Popstar what issue Don?
[11:23] Leighton.Marjoram acronym soup my favourite
[11:23] Marcus.Llewellyn Where are IMA meeting locations posted, Selby? I subscribed to the IMA mailing list and calendar, but neither seems to be used. So I never know where the next meeting will be.
[11:23] Don.Smith the issue : when applying a texture to multipule objects < prims > at the same time, the texture will not stick to all surfaces, sometimes not even after clicking to reapply multipule times
[11:24] Don.Smith sometimes not at all
[11:24] Arielle.Popstar ahh ok that was an issue raised last week?
[11:24] Don.Smith Selby here has been investigating and testing < with help from a Kayaker test object > to help me find if a solution can be found and implimented to fix this
[11:25] Selby.Evans I will see Marcus--I get the info from Steve -- but there should be more promotion, I guess -- This friday the meeting will be at my place on Kitely
[11:25] Marcus.Llewellyn Are these prims linked? Or simply all selected?
[11:25] Marcus.Llewellyn Thanks Selby. :)
[11:25] Don.Smith just selected, being linked doesnt seem to affect the problem < meaning even if linked same issue >
[11:26] Selby.Evans the Prims are linked, Marcus
[11:26] Arielle.Popstar ok was raised by Selby's testing. Ubit mentioned something about it to me
[11:26] Don.Smith Selbys test object they are linked, i have also done lots of testing usually with just selected prims
[11:26] Kayaker Magic: It can happen with a single prim, but having more faces by linking a bunch of prims together increases the probability of seeing one face fail to get the texture.
[11:27] Don.Smith Its enough of a problem for me when building it breaks my focus < which is why i build >
[11:28] Marcus.Llewellyn How many prims do you usually select at once?
[11:28] Selby.Evans with my test object, I get a fail in 12 tests. in most grids
[11:29] Don.Smith as little as 5 and up to 100+ for my testing
[11:29] Don.Smith can happen on a single prim
[11:29] Marcus.Llewellyn Ah.... there. Just happened to these prims.
[11:29] Marcus.Llewellyn One didn't take.
[11:29] James.atLLOUD Prims have an 'order' when multi-selected, no?
[11:30] Bill.Blight I'm still convinced this issue is as much a networking issue as an opensim issue
[11:30] Don.Smith i am.....lets just say, an Insane builder, i build faster then most people and often i design a build and try to texture in groups of surfaces at a time, this issue causes me to keep stopping and fixing textures, breaking my focus on design and building further
[11:30] Marcus.Llewellyn How very random... some random face just doesn't seem to get the update
[11:30] Arielle.Popstar shouldnt be an issue then on localhost sim
[11:31] Bill.Blight You can almost watch the packets drop when this occurs
[11:31] Leighton.Marjoram I can see how that would be distracting @Don
[11:31] Bill.Blight I never have this issue, which is weird ..
[11:31] Gavin.Hird any difference between viewers?
[11:31] Don.Smith here is a 5000 prim Ballroom/shopping mall i designed and built in 12 days, its got a glass dance floor over a massive waterfall all set on the ocean.
[11:32] Leighton.Marjoram i've only noticed since people starting talking about #SelectiveAttention
[11:32] Don.Smith this video is the grand opening event there ..
[11:32] Don.Smith viewer seems to have zero effect on this issue
[11:32] Bill.Blight I tested with Firestorm 64 , kokua and alchemy 5 beta and can't make it happen , but if I go to a remote sim, external to my grid, I can make it happen all the time .. like to a sandbox here on OSGRID ...
[11:33] Gavin.Hird did you try to adjust the setting in OpenSim.ini how often changes are flushed to disk?
[11:33] Arielle.Popstar what about changing the bandwidth limit?
[11:33] Don.Smith I've been on kitely, so i dont think i can adjust that < can i? >
[11:33] Arielle.Popstar of the viewer
[11:34] Gavin.Hird no you can't
[11:34] Don.Smith bandwidth limit i have tried changing, i got no change in the problem
[11:34] Bill.Blight Especially if you are running on a debian flavor of linux, the networking is anemic, there are some udp tweaks you can make that seem to make this issue vanish on local sims
[11:34] Bill.Blight without ever touching opensim code
[11:34] Arielle.Popstar Ubit did some work on udp recently
[11:35] Selby.Evans I tested bandwidth changes without success
[11:35] Gavin.Hird In this case the bandwidth does not matter much as it is just an update of xml that already exist in the asset server
[11:35] Gavin.Hird unless the linked mesh is massive
[11:35] Gavin.Hird or prims
[11:35] Bill.Blight Bandwidth does matter, if it is dropping udp packets
[11:35] Bill.Blight the viewer has to send the update to the region
[11:36] Don.Smith i believe you, however, when adjusting the bandwidth lower or higher, i got same results on kitely
[11:36] Bill.Blight well not bandwidth, congestion is more apt
[11:36] Selby.Evans Fred B. also suggested bandwidth changes. But I found no settings that fixed the problem.
[11:36] Gavin.Hird I'll test it with a completely different database and see what happens
[11:37] Arielle.Popstar anyone tested on master within the past couple weeks?
[11:37] Bill.Blight I run nothing but master
[11:37] Bill.Blight I posted a video not long ago, and can't make it happen
[11:37] Don.Smith Digiworlds has solved the issue, their server version of OS doesnt have this issue
[11:38] Marcus.Llewellyn It was happening for me here, although HG is possibly a factor.
[11:38] James.atLLOUD related?
[11:39] Marcus.Llewellyn Look like an unrelated mantis to me, James.
[11:39] James atLLOUD: ok :)
[11:40] Don.Smith I appreciate everyones inputs and thoughts on the issue
[11:40] Selby.Evans Apparently solved in OS master. Terry Ford does not know what he did to fix it.
[11:41] Arielle Popstar: Ubit did do something very recently in an attempt to resolve it
[11:41] Arielle Popstar: after Selby posted his tests
[11:42] Arielle Popstar: was there a difference between windows and linux servers btw?
[11:42] Selby.Evans I think Ubit did fix it -- And I hope he can give a copy of the patch to Kitely or to me
[11:43] Gavin.Hird Selby all changes are on the git for anyone to pick up
[11:44] Selby.Evans Arielle -- I did not keep track of the servers.
[11:44] Bill.Blight The fix was for the .9x codebase, so not sure it is gonna help any grid not up to speed on that
[11:44] Selby.Evans Oh -- thanks Gavin
[11:44] Gavin.Hird Kitely will potentially have to re-work it for their code base
[11:45] Arielle Popstar: i would say we need a .9 release
[11:45] Sheera.Khan they have a core dev of their own: Oren ...
[11:45] Selby.Evans Yes, Gavin, but that puts the ball in their court.
[11:45] Leighton.Marjoram Agreed Arielle
[11:45] Sheera.Khan so there's one familiar with the code base
[11:46] Bill.Blight Can't say I blame the dev team for not doing a release, with all the complaining about this and that, and the mesh , yada yada yada
[11:46] Marcus.Llewellyn Well, Oren is great, but so much code has been refactored/added/changed with 0.9. It's a lot to keep up with.
[11:46] Arielle Popstar: shrugs, we always complain so nothing new in that
[11:47] Don.Smith Orin and Ilan are currently working on a big project and i was informed they did not have the time to look into my problem, even if this change/fix was given to them, i'd be surprised if they fixed it for me tbh
[11:47] Bill.Blight That is true, but the vile stuff being said about .9 goes even above that ..
[11:47] Arielle Popstar: they have gotten over it for 10 years so why not now
[11:47] Arielle Popstar: who is saying it?
[11:47] Arielle Popstar: the ones on it like it
[11:47] Gavin.Hird I think perhaps the 0.9 codebase has to settle down a bit and be tested unchanged for a while before release to a wider audience
[11:47] Gavin.Hird too many changes going on right now
[11:48] Bill.Blight I would agree with Gavin on that
[11:48] Selby.Evans Ilan asked me to look for a fix and now I have one.
[11:48] Bill.Blight I run bleeding edge, and my users are happy with it, but they understand that it is a tad bit "different"
[11:48] Marcus.Llewellyn There's always people who resist an upgrade. Sometimes for trivial/subjective reasons, but grid operators do have legitimate concerns about going to a new version.
[11:48] Arielle Popstar: i think testing by a wider audience would be good too
[11:49] Gavin.Hird it depends Arielle
[11:49] Arielle Popstar: whats the biggest issue with .9 by those who havent upgraded?
[11:49] Gavin.Hird we don't want a repeat of the mesh disaster that struck for the 0.8.2 release
[11:49] Bill.Blight the biggest thing people complain about is the mesh issue with ubODE, but they can still run bullet if they want
[11:49] Marcus.Llewellyn It's trivial to keep using Bullet.
[11:50] Arielle Popstar: but can stay on bullet
[11:50] Arielle Popstar: so not really a legit complaint
[11:50] Bill.Blight .9x is a little bit more of a memory hog
[11:50] Arielle Popstar: imo
[11:50] Bill.Blight but not much
[11:50] Marcus.Llewellyn It also wouldn't hurt if more grid owners ran test regions with things like ubODE enabled, so that they can contribute rather than complain.
[11:50] Don.Smith ok, i am from the future < lol > and i have decided to give you each mental powers with the ability to create alternate realities where you can do anything you can imagine without effect on real life. You're all Welcome! < As if i could rofl > Call it Open Sim 1million
[11:51] Bill.Blight I run ubODE on all my .9x regions
[11:51] Andrew Hellershanks: Sorry I'm late. I distracted myself by trying to finish off something before I came in. When I did try to get in I found I was having trouble.
[11:51] Leighton.Marjoram me too @Bill
[11:51] Gavin.Hird I run some with Bullet and some with uBODE
[11:51] Arielle Popstar: yes Marcus on more grids testing
[11:51] Bill.Blight and I have done lots of tests and videos, but people seem to gloss over them .. Because I'm not a big guy
[11:51] Arielle Popstar: am curious how many havent upgraded
[11:51] Bill.Blight LOL
[11:51] Arielle Popstar: Metro and Kitely
[11:52] Arielle Popstar: what other big grids?
[11:52] James atLLOUD: with each release, is one physics engine the 'recommended' one?
[11:52] Arielle Popstar: usually a default one but can be changed per grid
[11:52] Andrew Hellershanks: Yes. Change in physics engine was a big bone of contention for a lot of people.
[11:52] Marcus.Llewellyn Some aren't big per se, but still popular. 3rd Rock Grid, Great Canadian Grid, Francogrid are examples.
[11:52] Sheera.Khan Metro is on the latest "released" OpenSim but allows newer versions to be connected
[11:53] Arielle Popstar: any complaints from those on newer Sheera?
[11:53] Selby.Evans Marcus, would it help if the IMA computer ran some test regions like that?
[11:53] Sheera.Khan not to my knowledge...
[11:53] Sheera.Khan most complaints are results of other problems
[11:53] Arielle Popstar: nod
[11:54] Marcus.Llewellyn As long as the IMA grid admins also submit mantis reports, it certainly would, I should think, Selby. :)
[11:54] Sheera.Khan like IPv6...
[11:54] Arielle Popstar: so Andrew, whats still needed to make a .9 release?
[11:54] Gavin.Hird if Robust is not updated there could be all kinds of issues on simulators running 0.9
[11:54] Gavin.Hird and that swings both ways
[11:55] Bill.Blight If the IMA is going to test it, and release "results" they need to make sure they have someone who fully understands .9x before just throwing a region up and testing ..
[11:55] Sheera.Khan up to now we didn't experience problems with our 0.8.2+ Robust
[11:55] Sheera.Khan even though there are quite some 0.9 regions
[11:55] Andrew Hellershanks: Arielle, I don't know. You could check the list of mantis reports to see what bugs are still open. I haven't talked with the other core devs lately to know what still needs to be done.
[11:55] Marcus.Llewellyn That's a good point. Occasionally, simulators do require the newer ROBUST, making running release and master side by side more challenging.
[11:56] Arielle Popstar: before the stated holdup was permissions but we no longer hear of complaints there
[11:56] Gavin.Hird I have experienced all kinds of oddities if robust and simulators are not on more or less the same 0.9 dev version
[11:56] Bill.Blight I run a .9x robust but have .8x connected regions as well, and have not run into any actual issues
[11:57] Andrew Hellershanks: Marcus, what do you mean by "some aren't big per se"? What is "big". AFAIK, The GCG has about 1,000 regions online. You don't consider that big?
[11:57] Marcus.Llewellyn Yeah, I can't think of anything I've heard about that should break sim compatibility with ROBUST. Doesn't mean there aren't a few edge cases, though. ;)
[11:58] Bill.Blight you may see some debug messages that either .8x does not understand from .9x or vice versa but no actual issues
[11:58] Marcus.Llewellyn I actually didn't know GCG was that big! Hehe. But was meaning grids that float around the 50-100 region mark. There's many of them.
[11:58] Arielle Popstar: most of the bigger grids are fairly modified from stock opensim
[11:58] Andrew Hellershanks: Marcus, I don't know how they did it but they got big very quickly.
[11:58] Bill.Blight that is true, even I run with a couple of mods, but nothing that really impacts the core
[11:58] Arielle Popstar: i thought they dropped again
[11:58] Andrew Hellershanks: Marcus, They capped the number of regions at 1,000 IIRC.
[11:59] Arielle Popstar: but no idea of their current as they no longer release their figures publically
[11:59] Marcus.Llewellyn Good on GCG. :)
[12:00] Don.Smith All of your work and efforts to improve OS are appreciated guys/gals!
[12:02] Andrew Hellershanks: Too bad misterblue isn't at these latest meetings.
[12:02] Arielle Popstar: scary question, do the core devs still talk to each other?
[12:03] Andrew Hellershanks: Arille, yes we do.
[12:03] James atLLOUD: IRC IIRC - lol
[12:03] Arielle Popstar: ok you said you hadnt talked with any recently
[12:03] Arielle Popstar: so wondered
[12:03] Marcus.Llewellyn We don't always see all of their communications. They have their own private chats.
[12:04] Andrew Hellershanks: Arielle, not specifically about what needs to be done in order to release 0.9
[12:04] Andrew Hellershanks: There are 5 core devs in our core IRC channel right now.
[12:04] Arielle Popstar: to my mind it is the big question
[12:04] Arielle Popstar: when and what needs to be done to get it there
[12:05] Andrew Hellershanks: Arielle, I'll get that discussion going amongst the core group
[12:05] Arielle Popstar: kk that would be great Andrew
[12:06] Bill.Blight did Ubit bring up the osNotecard threat levels to you guys?
[12:06] Bill.Blight I am still trying to figure out why those functions are such a hight threat level
[12:06] Marcus.Llewellyn What about those threat levels?
[12:06] Andrew Hellershanks: I did see a question about why it has the threat level it does have. I didn't see any response to that question
[12:07] Bill.Blight you can do those functions with lsl but you have to use the dataserver and it is slow and a lot more code ..
[12:07] Sheera.Khan That discussion comes up from time to time...
[12:07] Sheera.Khan I remember it being discussed and attibuted to DoS attacks
[12:07] Marcus.Llewellyn They're as they are because they can potentially put a very high load on the asset service(s).
[12:07] Bill.Blight potentially so can rezobjecty
[12:07] Bill.Blight LOL
[12:07] Arielle Popstar: hehe
[12:07] Andrew Hellershanks: One of the reason a function can get a higher threat level than one might think has to do with what takes places in OpenSim to execute the function.
[12:07] Marcus.Llewellyn Believe me, I WISH they could have a lower threat level and be more usable. Those are wonderful functions. But it would be trivial for griefers to misuse them.
[12:08] Andrew Hellershanks: I've asked about the threat level of a few functions in the past when I was adding them.
[12:08] Arielle Popstar: per region, grid services or database?
[12:08] Bill.Blight there are so many other functions that you can bring a sim to its knees if you have scripting rights, if they are gonna do it they'll find a way
[12:08] Andrew Hellershanks nods to Bill
[12:09] Marcus.Llewellyn True, but osNotecard functions are, as I said, trivial to abuse, even unintentionally. A large number of users with HUDs that use updateable notecards could even be an issue.
[12:10] Bill.Blight hell you can bring a sim to it's knees with chat spam, and don't need anything but basic rights for that ..
[12:10] Marcus.Llewellyn There's nothing stopping a grid owner from making them generally available. But on the whole, they're safer as a high threat level. :(
[12:10] Andrew Hellershanks: Yes, Marcus. That is one of the reasons that the os function to create notecards is restricted by default. A lot of script authors saw the function listed and assumed they could (or should) be allowed to use them.
[12:11] Bill.Blight well , I'll say this, I can bring a region to it's knees with using dataserver events to read notecards, so I don't really see the difference ...
[12:11] Andrew Hellershanks: The threat levels can be a guide to grid owners to think about what might happen if some functions were enabled and whether they feel they would be safe to have enabled in their grid.
[12:12] Arielle Popstar: how much problem does it create with not being able to use it
[12:12] Andrew Hellershanks: In other words, they need to think about what is being enabled before just going ahead and enabling a ton of stuff without thinking about the implications.
[12:13] Bill.Blight Well I'll give you this example of use, I made a tp hud that reads it's locations from a notecard .. works great .. BAM land on a foreign sim SCRIPT ERROR
[12:13] Arielle Popstar: right
[12:13] Andrew Hellershanks: Arielle, a lot of problems when you have to deal with a lot of users and/or script writers who have gadgets that want to create notecards and they find it isn't permitted in the grid.
[12:13] Arielle Popstar: so its a pain
[12:13] Arielle Popstar: with a high threat level
[12:13] Marcus.Llewellyn It doesn't create a problem.. But it does mean we can't do more nifty things. For example. many objects store settings only until a script is reset. the notecard functions could allow a creator to write the configuration to a notecard, and the settings would persist, or even be interchangable.
[12:13] Andrew Hellershanks: Bill, if you are just reading locations from an NC there is no real need to do that using os functions.
[12:13] Tannoy OSgrider: I think that at least reading notecard could be at lower thread lvl
[12:13] Bill.Blight My specific question to Ubit was osGetnotecard functions
[12:14] Marcus.Llewellyn The problem is that everytime you write to a notecard, even a single character, you're creating a whole new asset.
[12:14] Bill.Blight I can understand the make functions being higher
[12:14] Andrew Hellershanks: For some script writers it is a problem because they may want to sell their items in multiple grids without having to have different versions for different grids.
[12:15] Bill.Blight The osGetNotcardX functions , IMHO, should not be that high
[12:15] Arielle Popstar: so it is possible to break a grid or a database with it?
[12:15] Marcus.Llewellyn I actually do just that. I have a INI style configuration script that reads from notecards.
[12:15] Marcus.Llewellyn There's also a version that can write to them.
[12:15] James atLLOUD: what would be an alternative to a notecard to store customizations for a script?
[12:15] Andrew Hellershanks: Bill, That is a question for Ubit and/or Melanie who would have a better idea of what is going on in the background when using that function.
[12:15] Bill.Blight using the LSL Dataserver and LSL functions to read notecards
[12:16] Marcus.Llewellyn the usual alternative is to use something offsite using HTTP. a keyvalue store for example.
[12:16] Bill.Blight those are unrestricted
[12:16] James atLLOUD: ah ty Marcus
[12:16] Arielle Popstar: that raises the level of complexity
[12:16] Andrew Hellershanks: Marcus, yes that can be done but a lot of people wouldn't know how to do it or wouldn't have access to an offsite database.
[12:16] Marcus.Llewellyn Agreed. I don't consider it ideal.
[12:16] Bill.Blight You can use the LSL dataserver, but there is a built in delay, it is slow and does not do really well with multiple quick calls ...
[12:17] Andrew Hellershanks: Arielle, it does a bit but not too much.
[12:17] Sheera.Khan and it would help if one could safely delete objects from the asset storage
[12:17] Marcus.Llewellyn What would be nice are mutable assets.
[12:17] Bill.Blight warning chat spam incoming
[12:17] Bill.Blight dataserver(key query_id, string data)
if (query_id == notecardQueryId)
if (data == EOF){
// llOwnerSay("Done reading notecard, read " + (string) notecardLine + " notecard lines.");
// llOwnerSay((string)tmplist);
Dlist = tmplist;
Llist = llList2ListStrided(llDeleteSubList(Dlist, 0, 0), 0, -1, 2);
Dlist = llList2ListStrided(Dlist,0,-1,2);
// llOwnerSay((string)Llist);
// llOwnerSay((string)Dlist);
// bump line number for reporting purposes and in preparation for reading next line

tmplist = tmplist + data;
//llOwnerSay( "Line: " + (string) notecardLine + " " + data);
notecardQueryId = llGetNotecardLine(notecardName, notecardLine);
[12:18] Andrew Hellershanks: If you are using it for configuration options they should only need to be read once when the item is first rezzed.
[12:18] Bill.Blight example of using the data server to read notecards, but can be done with about 4 lines with os functions
[12:19] Andrew Hellershanks: If an item is rezzed, configured, and taken back to inventory the settings will stay when the item is rezzed again. The only problem is if the item gets reset as that would make the script lose any changes to its default configuration
[12:19] James atLLOUD: the threat is that any scripter can abuse reading a notecard, yes?
[12:20] Marcus.Llewellyn How difficult would it be to create a mutable asset type?
[12:20] Bill.Blight but any scripter CAN read a notecard without OS functions
[12:20] Andrew Hellershanks: James, yes they could.
[12:20] Bill.Blight the threat is in MAKING notecards with the os functions
[12:20] Marcus.Llewellyn James, on a lower level, they could give the sim cache or grid asset service a real beating.
[12:20] James atLLOUD: yes. thanks.
[12:21] Bill.Blight Reading notecards should be a NON-issue, for os functions, making could be an issue
[12:21] Andrew Hellershanks: Bill, Yes, a scripter can read NC's without using os functions.
[12:21] Sheera.Khan well, the dataserver event is throttled to reduce the potential for attacks
[12:21] Bill.Blight yes but you can still kill a region with it
[12:21] Andrew Hellershanks: Once you have a script that reads an NC it is easy to reuse that code in other scripts.
[12:22] Bill.Blight I still don't see how reading is a threat ... <shrug>
[12:22] Marcus.Llewellyn Sadly, we can't presume that all scripts will behave rationally or responsibly.
[12:22] Tannoy OSgrider: dataserver is higher threat than reading notecard
[12:22] Bill.Blight I'd think so
[12:22] Andrew Hellershanks: Bill, You would have to understand what goes on behind the scenes in order to know why a function was given the threat level it has been assigned.
[12:23] Bill.Blight You assume I don't .. LOL ... . :P
[12:23] Bill.Blight I understand why making notecards is that threat level , believe me
[12:23] Andrew Hellershanks: Bill, I'm not saying that you in particular don't.
[12:23] Bill.Blight I was just being cheeky
[12:24] James atLLOUD: :)
[12:24] Andrew Hellershanks: :)
[12:24] Bill.Blight Like I said, don't really have an argument with the make notecard function
[12:24] Andrew Hellershanks: I thought a function I added to the os functions was a low threat until I was told why it needed a higher threat level.
[12:24] Bill.Blight but even though the dataserver is throttled, it can balloon to kill a region
[12:25] Andrew Hellershanks: No, but it is a good question as to why the os function to read an NC has the threat level it has.
[12:25] Marcus.Llewellyn As I said before, I adore the notecard functions, They're the cat's meow. But in order to convince devs that they're not a threat, you're going to need to show them real performance testing with numbers. Something definitely empirical.
[12:25] Arielle Popstar: unfortunately the reasonings arent made known to the people who would benefit from the knowledge
[12:25] Bill.Blight I guess my thought is , OS functions were supposed to be an improvement to LSL functions, but in the readnotecard instance it is a downgrade ...
[12:26] Andrew Hellershanks: Making the reasons known could benefit people, including the griefers who might then known better ways to mess up a region or grid.
[12:26] Marcus.Llewellyn I asked about the notecard functions years ago. Melanie explained it to me in no uncertain terms that they can be an issue. And I ain't arguing with someone who knows OpenSim as well as Mel does. ;)
[12:27] Bill.Blight ok empirical data, 40 lines of code versus 4, 1.5 ms delay per line compared to .5
[12:27] Arielle Popstar: Mel does have a high security level bias
[12:27] Andrew Hellershanks: Some functions can require database access in order to do what they need to do. That can result in a function having a higher threat level than some other function.
[12:27] Bill.Blight so you are adding not only to the timer, but to the script compile engine
[12:28] Arielle Popstar: that other grid operators may not share if they understand the risks
[12:28] Andrew Hellershanks: Arielle, Any grid owner should have a high security bias.
[12:28] Bill.Blight yes
[12:28] Bill.Blight I agree
[12:28] Marcus.Llewellyn Just because your script has fewer lines does not mean that the implementation behind that function is a simple one. That's not a logical assumption.
[12:28] Arielle Popstar: a commercial grid perhaps
[12:28] Arielle Popstar: many arent
[12:28] Bill.Blight I just don't understand the read threat level, the others I understand
[12:29] Bill.Blight Again I was just being cheeky Marcus
[12:29] Andrew Hellershanks: Bill, how did you come up with the delay times?
[12:29] Marcus.Llewellyn Sorry, I missed the nuance. :)
[12:29] Bill.Blight You can look at the lsl api and see the built in delays
[12:29] Bill.Blight and I tested by spitting out wall clock at each line
[12:30] Andrew Hellershanks: Is that delay really going to be important if you are only reading the data when an item is rezzed (for example)?
[12:30] Bill.Blight depends on how many lines
[12:30] Tannoy OSgrider: than we can put down and textures loading on high treat because it's accessing grids db
[12:31] Bill.Blight If you tell me the DB cannot handle calls for text from a notecard, I'm gonna tell you to fix the db not the script function ..
[12:31] Bill.Blight :P
[12:31] Arielle Popstar: we need to get Mel to a few of these meetings
[12:31] Andrew Hellershanks: Tannoy, they might have higher levels. I still haven't finish updating the OS wiki with the threat level and script delays for os functions
[12:32] Tannoy OSgrider: I know, I'm only kidding about this
[12:32] Andrew Hellershanks: It just reminds me of yet one more item on my todo list.
[12:33] Bill.Blight of course on my grid NOW all the LSL calls built in timings are 10% of what they were, and I have had no issues ... Yes I modded the api
[12:33] Andrew Hellershanks: Arielle, we also need misterblue here if we are going to keep talking about physics engine.
[12:33] Arielle Popstar: last he mentioned was about needing to rewrite Bullet
[12:34] Bill.Blight To me it appears that some aspects of the code are still considering 10-15 year old hardware and speeds ...
[12:35] Andrew Hellershanks: Bill, how did you manage to get those script times down so much and are you planning on sharing the changes?
[12:35] Bill.Blight If a modern MySQL server cannot handle thousands of calls a second, then there is an issue with the DB not the core code
[12:35] Bill.Blight they are all in the LSL api
[12:35] Bill.Blight they are pretty much right at the top
[12:35] Bill.Blight I'll grab an example
[12:36] Bill.Blight in the public class LSL_Api : MarshalByRefObject, ILSL_Api, IScriptApi
[12:36] Bill.Blight section of the LSL_Api
[12:37] Bill.Blight this is unmoddified code btw
[12:37] Bill.Blight protected int EMAIL_PAUSE_TIME = 20; // documented delay value for smtp.
protected int m_sleepMsOnSetTexture = 200;
protected int m_sleepMsOnSetLinkTexture = 200;
protected int m_sleepMsOnScaleTexture = 200;
protected int m_sleepMsOnOffsetTexture = 200;
protected int m_sleepMsOnRotateTexture = 200;
protected int m_sleepMsOnSetPos = 200;
protected int m_sleepMsOnSetRot = 200;
protected int m_sleepMsOnSetLocalRot = 200;
protected int m_sleepMsOnPreloadSound = 1000;
protected int m_sleepMsOnMakeExplosion = 100;
protected int m_sleepMsOnMakeFountain = 100;
protected int m_sleepMsOnMakeSmoke = 100;
[12:37] Bill.Blight some of them are crasy
[12:37] Bill.Blight protected int m_sleepMsOnGiveInventory = 3000;
[12:37] Bill.Blight protected int m_sleepMsOnLoadURL = 10000;
protected int m_sleepMsOnParcelMediaCommandList = 2000;
protected int m_sleepMsOnParcelMediaQuery = 2000;
protected int m_sleepMsOnModPow = 1000;
protected int m_sleepMsOnSetPrimURL = 2000;
protected int m_sleepMsOnRefreshPrimURL = 20000;
[12:38] Bill.Blight those are stock numbers not the ones I use
[12:38] Bill.Blight I reduced all of them to 10% of their settings
[12:39] Andrew Hellershanks: Without checking I think those numbers are the delays imposed on the functions according to SL.
[12:39] Bill.Blight yep
[12:39] Bill.Blight but we not SL
[12:39] Bill.Blight and I am sure those numbers were for 10-15 year old viewers and server hardware
[12:40] Andrew Hellershanks: No, but you have to decide whether reducing those penalties on some functions will make it more likely that they could be abused.
[12:40] Bill.Blight you can see a difference in how fast scripts work , in a big script that has multiple functions ..
[12:40] Andrew Hellershanks: How has reducing those times impacted the system load?
[12:41] Bill.Blight Load , NONE
[12:41] Bill.Blight but then again, my servers are NOT overloaded , as I prefer to have more hardware than I need
[12:41] Tannoy OSgrider: I'm interested how much is loading animations in this
[12:41] Andrew Hellershanks: If you think those limits are slowing down scripts why did you just reduce the values to 10% of their original instead of just setting them to 0?
[12:45] Andrew Hellershanks: Anything else before we wrap it up for today?
[12:46] Andrew Hellershanks: Lowering some of the limits can have an impact on the script runtime and how it might impact other scripts and/or the region.
[12:47] Bill.Blight reducing to 10% was the easiest .. just took a zero off the end, wanted to live dangerously not deadly ..
[12:47] Sheera.Khan sure it would have an impact...
[12:47] Sheera.Khan that's what we want to see at times^^
[12:47] Andrew Hellershanks: Bill, :)
[12:48] Bill.Blight I have a lot of heavy scripts, and I have not seen a degradation of performance, but I do notice an increase in script response time
[12:48] Andrew Hellershanks: Using delays can have an impact so removing the need for them could help in several ways. Whether it makes some functions more open to abuse by lowering the delay times would depend on the particular function.
[12:49] Andrew Hellershanks: Bill, it takes longer for some scripts to respond?
[12:49] Bill.Blight does not really matter to me if the script takes 1 minute or 3 minutes to crash the sim the result ends up being the same .. Since you usuallly don't know till it goes down anyway
[12:50] Andrew Hellershanks: A couple of the other core devs never seem to be available at this time of day.
[12:51] Sheera.Khan :-(
[12:51] Sheera.Khan maybe we could move the event?
[12:51] Bill.Blight When you have a LARGE .. let's say 500 line script, that uses a lot of those functions, there is a noticeable difference between mine and stock
[12:51] Bill.Blight LOL
[12:51] Andrew Hellershanks: I've asked a couple of questions and no response yet.
[12:51] Bill.Blight george just landed on my sim
[12:51] Bill.Blight LOL
[12:51] Sheera.Khan it seems the sessions got more constructive lately
[12:51] Bill.Blight err grid
[12:52] Andrew Hellershanks: Bill, 500 lines is not large. I have a script file that is 3,000 lines long.
[12:52] Bill.Blight so do I
[12:52] Andrew Hellershanks: :)
[12:52] Bill.Blight but around 500 is wher I start noticing a difference
[12:52] Andrew Hellershanks: Sheera, good.  :)
[12:53] Bill.Blight Ubit knows about my script timings
[12:53] Bill.Blight lol
[12:53] Andrew Hellershanks: Bill, ok. I can talk about them with him next time he is showing signs of life in IRC
[12:53] Bill.Blight I mainly started doing them to speed up vehicle scripts
[12:54] Andrew Hellershanks: Did you notice the script runtime numbers go up, down, or stay the same?
[12:54] Bill.Blight and as Ubit has driven my vehicles , badly I may say, he has seen it ..
[12:54] Bill.Blight the runtime numbers bounce more
[12:54] Andrew Hellershanks: ok
[12:54] Bill.Blight but not any higher than I would say is normal
[12:54] Andrew Hellershanks: On average are they about the same as before your changes?
[12:54] Bill.Blight correct
[12:54] Andrew Hellershanks: ok
[12:55] Andrew Hellershanks: Ah. Mel has started answering my comments
[12:55] Bill.Blight You are welcome to come over and see for yourself at any time ..
[12:56] Bill.Blight Ubit has been using me for a crash test dummy for a while ...
[12:56] Andrew Hellershanks: Mel said the defaults will remain the same as SL. They can be changed if someone wants to change them.
[12:56] Andrew Hellershanks: They are settable.
[12:57] Bill.Blight they are not settable .. LOL .. unless you mod the code before compiling
[12:57] Sheera.Khan smells a new ini-file...
[12:57] Bill.Blight yep
[12:57] Sheera.Khan Delays.ini
[12:57] Bill.Blight what I was gonna say
[12:58] Bill.Blight and they are "protected" settings in the code, so don't think they are open at current to being changed via an ini
[12:58] Andrew Hellershanks: She said people who change the delays open them up to all kinds of overall performance issues even if it makes their pet script run faster. The script delays are there in order to make it impossible for one script to get all the sim's power or to guard the central services from flooding.
[12:59] Bill.Blight hogwash , what is she running on a 286
[12:59] Bill.Blight LOL
[12:59] Bill.Blight :P
[12:59] Andrew Hellershanks: I'm not sure if she meant settable by ini file or just that they were pulled out to near the top of the file.
[12:59] Bill.Blight I have sims with 3000+ scripts and yet to overload a sim
[13:00] Bill.Blight I can understand her point, but at the same time, let's grow up some ..
[13:00] Andrew Hellershanks: Sure. np when you are using a single well behaved script. What happens when someone puts one of those in a loop.
[13:00] Bill.Blight I screw up and put scripts in loops all the time .. who hasn't , they are lying
[13:01] Andrew Hellershanks: The os function to read a notecard can cause problems with the asset server.
[13:01] Bill.Blight then fix the asset server, don't limit the function
[13:01] Bill.Blight LOL
[13:01] Bill.Blight :P
[13:02] Andrew Hellershanks: If you know more than Mel go ahead and tell her that.
[13:02] Bill.Blight I guess I agree with some that say , we need to quit taking our example from SL and pass them by
[13:02] Andrew Hellershanks: I'm just passing along some of what she is telling me and she knows how the OpenSim code works.
[13:02] Bill.Blight I am not going to say I know MORE than Mel, not gonna say I'm stupid either ...
[13:03] Andrew Hellershanks: If you think it is safe to reduce limits you can try persuading her it is safe to do so. I think she will simply say it isn't.
[13:03] Bill.Blight But some people get stuck in a rut, and it happens to devs too ...
[13:03] Bill.Blight Also some people just hate for ideas to come from outside themselves
[13:03] Andrew Hellershanks: The defaults are also set to keep your average person using the code "safe".
[13:04] Bill.Blight Not worth the argument
[13:04] Bill.Blight I have a patch I apply before compiling
[13:04] Andrew Hellershanks: Whatever.
[13:04] Bill.Blight does me fine
[13:04] Andrew Hellershanks: You need to discuss it with Melanie.
[13:05] Sheera.Khan leading to situations where someone opens the gates of hades by setting the threat-level to severe
[13:05] Bill.Blight yeah
[13:05] Bill.Blight you can shoot yourself in the foot
[13:06] Andrew Hellershanks: We can't prevent people from shotting themself in the foot.
[13:06] Bill.Blight I am fine with protecting the masses, as long as it does not force people to use it the way other people want
[13:06] Andrew Hellershanks nods
[13:06] Bill.Blight beauty of opensource code
[13:06] Andrew Hellershanks: yup
[13:07] Bill.Blight I would never fork OS because I love it the way it is, but I'll throw in my .02 at times
[13:07] Bill.Blight LOL
[13:07] Bill.Blight I'm just personally patch happy, and I can live with that .. I don't like something, I know I can change it ..
[13:07] Andrew Hellershanks: Melanie has explained why the reading notecard function can kill things. I don't want to paste all that info here. I can IM it to you and you can feel free to disagree with Mel.
[13:08] Sheera.Khan well, but OTOH: those who run really time critical services or scrpts can edit the source code... - the average Jon Doe won't need some of those timing optimizations
[13:08] Sheera.Khan the OSGetNotecard maybe, but not the timings
[13:08] Bill.Blight I agree Sheera , but maybe it should be an ini setting
[13:11] Sheera.Khan maybe it would benefit the users more if we could tell them how to optimize their network settings
[13:11] Sheera.Khan I heard Debians handling of network traffic would be anemic...
[13:11] Bill.Blight yes it is
[13:12] Bill.Blight I posted the info at the last meeting I think
[13:12] Sheera.Khan there are quite some variables in /proc/sys/net one could tamper with^^
[13:12] Bill.Blight /etc/sysctl.conf
[13:12] Sheera.Khan sure :-)
[13:12] Sheera.Khan I even looked up some of them already
[13:13] Andrew Hellershanks: There are a lot of settings that can be adjusted on servers. Just another case of settings being certain values for the average use case I suspect.
[13:13] Bill.Blight let me grab mine again real quick
[13:13] Bill.Blight chat spam incoming
[13:13] Bill.Blight net.core.wmem_max=12582912
net.ipv4.tcp_rmem= 10240 87380 12582912
net.ipv4.tcp_wmem= 10240 87380 12582912
net.ipv4.tcp_window_scaling = 1
net.ipv4.tcp_timestamps = 1
net.ipv4.tcp_sack = 1
net.core.netdev_max_backlog = 5000
net.ipv4.tcp_keepalive_time = 3600
net.ipv4.udp_mem = 10240 87380 12582912
net.ipv4.udp_rmem_min = 32768
net.ipv4.udp_wmem_min = 32768

[13:14] Sheera.Khan wouldn't a tutorial on that be of help and the impact maybe really noticeable?
[13:15] Sheera.Khan ty, I'll compare them to my settings on CentOS - I'm rather fond of that one as it is quite stable although it is seldom challenged ...
[13:15] Andrew Hellershanks: Put it in a wiki page?
[13:15] Sheera.Khan might be a good idea...
[13:16] Andrew Hellershanks: It would be good to have an idea of the impact those settings may have on a machine and the network traffic.
[13:16] Sheera.Khan along with some recommendations regarding mono-version and -settings
[13:16] Andrew Hellershanks: Well that will wrap up this weeks meeting.
[13:18] Andrew Hellershanks: Those network settings might improve the throughput of the network as long as it doesn't choke the CPU.

Personal tools
About This Wiki