Chat log from the meeting on 2017-04-18

From OpenSimulator

Jump to: navigation, search

[11:00] Arielle Popstar: thx for the sign at wright Paela
[11:00] Marcus Llewellyn: Grrr... names aren't loading for me.
[11:01] Arielle Popstar: is Andrew on IRC? He used to go under Plugh didnt he
[11:03] paela argus: sorry for have moved the meeting but today wright is very laggy
[11:03] Nebadon Izumi: There are a lot of textures at Wright Plaza
[11:03] Nebadon Izumi: its probably around 5000 textures
[11:04] Arielle Popstar: this was a nice fast load
[11:04] George Equus: Cool place!
[11:04] Arielle Popstar: good view too
[11:04] Nebadon Izumi: I might have to leave in about 30 minutes
[11:04] Nebadon Izumi: i have to go pick up new Air Conditioner
[11:05] Nebadon Izumi: i went this morning only to get there and find out the hard way it wouldnt fit in my jeep
[11:05] Nebadon Izumi: was like 1 inch to big
[11:05] Nebadon Izumi: gigantic box
[11:05] Marcus Llewellyn: Argh. lol
[11:05] Gavin.Hird no resize handles?
[11:05] Nebadon Izumi: haha
[11:05] Nebadon Izumi: i figured if my 50 inch TV would fit
[11:05] Nebadon Izumi: so would the AC,
[11:05] Nebadon Izumi: but no go
[11:06] Nebadon Izumi: waiting on a ride, they are supposed to be here in like 30 minutes or so
[11:06] Arielle Popstar: not a portable one then
[11:06] Nebadon Izumi: its a Window Unit
[11:06] Nebadon Izumi: 15100 BTU
[11:06] paela argus: you can make it in plane go in france for me Neb :)
[11:06] Marcus Llewellyn: SUVs kinda suck that way. They advertise more space, but the reality of actually fitting things into that space is something else.
[11:06] Nebadon Izumi: the House AC just isnt enough with all these computers
[11:06] Nebadon Izumi: im on second floor
[11:06] Nebadon Izumi: and it gets quite oven like up here
[11:07] Nebadon Izumi: well its a Jacked up Jeep Wrangler Marcus
[11:07] Marcus Llewellyn: That's why I keep all my headless ones elsewhere ;)
[11:07] Arielle Popstar: server farms need good air conditioning :)
[11:07] Nebadon Izumi: its not the most Roomy vehicle
[11:07] Nebadon Izumi: big tires though :P
[11:08] Arielle Popstar: so what is happening in the world of opensim this week?
[11:08] Nebadon Izumi: Ubit and I were testing new ubODE last night
[11:08] Marcus Llewellyn: Bugs. :)
[11:09] Nebadon Izumi: ran into issues on the Raspberry Pi, but we got it fixed super late last night
[11:09] Nebadon Izumi: i need to compile new linux binaries for core still
[11:09] Nebadon Izumi: maybe later I will try to do that
[11:09] Nebadon Izumi: not sure if i can compile 32 bit binary on 64 bit or not, probably a switch to do that
[11:10] Arielle Popstar: that object TP is neat
[11:10] Nebadon Izumi: yea that is pretty cool
[11:10] Marcus Llewellyn: You can, althought I forget how to setup the toolchain.
[11:10] Nebadon Izumi: though dont try doing it to another sim just yet
[11:10] Nebadon Izumi: heh
[11:10] paela argus: solve the problem car mesh in mesh road ? :/
[11:10] Nebadon Izumi: at least while on a vehicle
[11:10] Arielle Popstar: i was helping him test it with 2 people in a vehicle
[11:10] Arielle Popstar: oh we did it between his sims
[11:10] Nebadon Izumi: maybe Paela, its supposed to help, I was to busy throwing prims around and it got late
[11:10] Arielle Popstar: but all on one instance
[11:11] Nebadon Izumi: it was wierd everything was fine on the 64 bit linux and windows
[11:11] Nebadon Izumi: but the raspberry pi went bonkers
[11:11] Nebadon Izumi: turns out he accidently deleted a couple lines of code he shouldnt have
[11:11] Marcus Llewellyn: Oops
[11:11] Nebadon Izumi: was odd though it only effected the Arm CPU
[11:11] Nebadon Izumi: and not the x86 ones
[11:11] Marcus Llewellyn: An endian thing, maybe?
[11:11] Nebadon Izumi: maybe
[11:11] Nebadon Izumi: OKC Raspberry Pi2 is up and running
[11:11] Nebadon Izumi: also Newton is running latest code
[11:12] Nebadon Izumi: prims no longer explode
[11:12] Nebadon Izumi: when you turn them physical and they are penetrating each other
[11:12] Nebadon Izumi: they used to violently explode
[11:12] Marcus Llewellyn: I'm still using Bullet. I haven't taken the time to figure out how ubODE does mesh physics.
[11:12] Nebadon Izumi: now they just kind of seperate nicely and fall
[11:12] Nebadon Izumi: its pretty much the same
[11:12] Nebadon Izumi: the thing people complain about
[11:13] Nebadon Izumi: is when a mesh object has no physical mesh uploaded with it
[11:13] Arielle Popstar: he mentioned that the bounce wasnt predictable though
[11:13] Nebadon Izumi: Bullet has the ability to use convex hull or the visible appearance
[11:13] Nebadon Izumi: as the physics mesh
[11:13] Marcus Llewellyn: Ah. Okay, good. As long as I can upload my own hull, I'm good.
[11:13] Nebadon Izumi: bullet does not
[11:13] Nebadon Izumi: so as long as you upload properly there is no difference
[11:13] Nebadon Izumi: the people complaining just really have bad objects
[11:13] Kayaker Magic: You mean ubOde does not?
[11:13] Nebadon Izumi: right ubODE doesnt do that
[11:14] Nebadon Izumi: so the object ends up being phantom
[11:14] Marcus Llewellyn: Might switch my sandbox to ubODE and finally play with it.
[11:14] Nebadon Izumi: the real fix is reupload your mesh correctly
[11:14] Nebadon Izumi: but some people just dont want to do that
[11:14] paela argus: never problem with my stuff mesh and ubode :)
[11:14] Nebadon Izumi: some people dont have the originals i guess
[11:14] Marcus Llewellyn: I'm not a vehicle guy, so those improvements don't really spark me.
[11:14] Nebadon Izumi: they are just using stuff someone else uploaded incorrectly
[11:15] Kayaker Magic: The lack of a physics shape was preventing some people from walking through the doors of their mesh houses.
[11:15] Marcus Llewellyn: Prolly neglected to provide a proper physics shape.
[11:15] Nebadon Izumi: yea
[11:15] Arielle Popstar: i thought that was to be the new norm?
[11:16] Marcus Llewellyn: Mesh creators need to take that seriously. Creating your own low poly hulls can really help sim performance.
[11:16] Nebadon Izumi: most of the problem stuff is super old
[11:17] Nebadon Izumi: I can say though if you are running opensim on a low power machine
[11:17] Nebadon Izumi: like the raspberry Pi
[11:17] Arielle Popstar: maybe possible to fix viewer side like the conversion of prims to mesh?
[11:17] Nebadon Izumi: there is a huuuuuge performance boost by switching to UbODE
[11:17] Marcus Llewellyn: GLWT.
[11:17] paela argus: i have seen ubode use more 15 % more cpu is not a problem on basic user computer with viewer run etc ?
[11:17] Nebadon Izumi: it uses a lot less CPU and memory
[11:17] Marcus Llewellyn: is ubODE multithreaded now?
[11:17] Nebadon Izumi: i went from using like 600-800mb of memory
[11:17] Nebadon Izumi: and like 80-120% cpu with single avatar
[11:18] Nebadon Izumi: to like 200-300mb and 20-40% cpu
[11:18] Nebadon Izumi: no its single thread
[11:18] Nebadon Izumi: but it doesnt matter
[11:18] Nebadon Izumi: its far better on low end machines
[11:18] Gavin.Hird if the original mesh is lost, you can always set it to phantom, and build a prim collission model that is set transparent
[11:18] Marcus Llewellyn: Does it at least run in it's own thread?
[11:18] Nebadon Izumi: most of the time with bullet even on single thread i was just about maxing the CPU out just standing there
[11:18] Nebadon Izumi: if i got on a vehicle i was seeing CPU use like 150-350%
[11:18] Nebadon Izumi: pretty much saturating the entire machine
[11:19] Nebadon Izumi: now with ubode we had 2 mars rovers going around it was fine
[11:19] Nebadon Izumi: wasnt even getting near 100% of single core
[11:19] Gavin.Hird The most important difference with uBode is that the collision model must be two sided. Some creators have only used planes, which work in Bullet
[11:19] Nebadon Izumi: on Raspberry Pi2 overclocked to 1ghz
[11:19] Nebadon Izumi: yes another big improvement with it is raycast
[11:19] Nebadon Izumi: its far better
[11:19] Nebadon Izumi: uses way less cpu
[11:20] Sheera Khan: Kay will love to hear that
[11:20] Nebadon Izumi: on my beefy machines im still mostly using Bulletsim
[11:20] Kayaker Magic: My ears perked up!
[11:20] Nebadon Izumi: but its nice to have an option on the really low power stuff
[11:20] Nebadon Izumi: good for VMs and Pi's
[11:20] Nebadon Izumi: there is some down sides to ubODE
[11:21] Nebadon Izumi: like the Torus Coaster doesnt work well
[11:21] Nebadon Izumi: Chains dont work well
[11:21] paela argus: we have test here on this region 68 avi with no lag on ubode a good improvement for opensim
[11:21] Nebadon Izumi: but most of that stuff is not stuff the average user is going to use
[11:21] Nebadon Izumi: especially on low end servers
[11:22] Nebadon Izumi: i'll try to get the new ubode files into core later today for linux users
[11:22] Nebadon Izumi: anyone running windows should be able to already test it I think
[11:22] Kayaker Magic: Neb, did you se my Mantis on the speed of llList fetching functions? And that it turns out only to be a problem on Windows servers?
[11:23] Nebadon Izumi: no I didnt see that
[11:23] paela argus: ⊙Ok ⊙
[11:23] paela argus: i'm open test on a server linux on osgrid neb :)
[11:23] Nebadon Izumi: how suprising
[11:23] Nebadon Izumi: usually its Linux that suffers those kind of performance issues
[11:23] Nebadon Izumi: tables are turning :P
[11:23] Kayaker Magic: ..looking up the mantis number...
[11:23] Arielle Popstar: mono will restore it no doubt
[11:24] Gavin.Hird Can someone get 8118 and 0805 ino core?
[11:24] Marcus Llewellyn: Yup, someone sure can. :)
[11:24] Gavin.Hird 08085*
[11:25] Nebadon Izumi: maybe later I can, im a bit scared about doing pgsql patches, since i have no way to test
[11:25] Nebadon Izumi: i assume they are well tested?
[11:25] Andrew.Hellershanks I wasn't sure where the meeting was being held. The TP didn't take me to here and there are two large groups of green dots in this region.
[11:25] Gavin.Hird They have been running steady for 3 months
[11:25] Nebadon Izumi: ok i made a note
[11:25] Nebadon Izumi: i will try to push them in later
[11:25] Andrew.Hellershanks Sorry I'm late. I was out of the house at a meeting. Only just got back.
[11:25] Gavin.Hird great
[11:26] Kayaker Magic: Mantis 8127 and it is significant, I have a program that runs in seconds on a linux server and HOURS on a Windows server.
[11:26] Nebadon Izumi: ouch
[11:27] Nebadon Izumi: yea probably best to see what Ubit and Melanie think about that one
[11:27] Nebadon Izumi: unfortunately im not much help with 8127
[11:28] Gavin.Hird what about 8147 - anyone else seen that or has it been tested properly on standalone?
[11:28] Nebadon Izumi: I havent run standalone in ages
[11:29] Gavin.Hird someone probably should from time to time :-))
[11:29] Gavin.Hird it is the same with me
[11:30] Nebadon Izumi: yea
[11:30] Isis.Ophelia Fred Beckhusen is working setting the Standalone for Dreamworld, maybe he could test it? just an idea
[11:30] Nebadon Izumi: every once and a while i fire one up so i can hack an OAR or something
[11:30] Arielle Popstar: thats windows
[11:30] Marcus Llewellyn: I can't remember the last time I ran a standalone. What mnodule is that?
[11:30] Nebadon Izumi: but that is pretty rare
[11:30] Nebadon Izumi: generally i run a super stock configuration
[11:30] Isis.Ophelia I am running his Dreamworld on windows 10 but
[11:30] Nebadon Izumi: like rename the example files and start it up
[11:30] Gavin.Hird it is the ETCd monitoring module
[11:31] Marcus Llewellyn: Well yes, I get that from the name. ;) What does it do/monitor?
[11:31] Gavin.Hird Melanie said in the commit it can also be run on standalone, but I suspect it has not been tested very well
[11:31] Marcus Llewellyn: Is that a stock module?
[11:32] Andrew Hellershanks: I haven't run in to the issue reported in 8147 and I run Standalone to do testing.
[11:32] Nebadon Izumi: yea unfortunately i have no experience with that module in grid mode or standalone
[11:32] Andrew Hellershanks: Marcus, I did find it in the OS source tree.
[11:32] Marcus Llewellyn: If it's not a core module, there's not much that can be done for it on OpenSim's mantis.
[11:33] Gavin.Hird there is this commit
[11:33] Andrew Hellershanks: It is under the part of the tree of OptionalModules.
[11:35] Andrew Hellershanks is reading the log of the meeting to this point...
[11:35] Kayaker Magic: Misterblue, what is the status of mantis 8010, where Bullet sometimes fails to send collision messages?
[11:35] Andrew Hellershanks: If ubODE is that much better on memory and CPU usage I will be looking at switching over to it on a grid.
[11:36] Gavin.Hird First reference to ETCD is in
[11:36] Misterblue Waves: I haven't looked at it in a while.... I couldn't decide if the physics engine was failing or if it was the propagation of the collision into the simulator
[11:37] Nebadon Izumi: Andrew yea just be sure to test it
[11:37] Nebadon Izumi: its not a perfect replacement
[11:37] Nebadon Izumi: but i think for like 75% of people it will be good enough
[11:37] Andrew Hellershanks: I've caught up on reading the meeting log
[11:37] Misterblue Waves: the Bullet physics engine hasn't been updated in a while (the base version 2 people are using)
[11:37] Nebadon Izumi: some vehicle scripts from bullet may not work exactly as they did though
[11:37] Misterblue Waves: Bullet3 has been slowly advancing (with GPU acceleration) but it's not ready for prime time
[11:37] Nebadon Izumi: so i would suggest a very thorough test before switching
[11:38] Andrew Hellershanks: nebadon, but you just know that the other 25% will be vocal about any problems. :)
[11:38] Arielle Popstar: lol Andrew
[11:38] Andrew Hellershanks: hm... llRayCast is much improved now in ubODE? I know that's been a concern of Kayaker.
[11:38] Nebadon Izumi: yes, its proper C++ ray
[11:38] Nebadon Izumi: not C#
[11:39] Andrew Hellershanks: Misterblue, good to hear things are progressing in B3
[11:39] Nebadon Izumi: its way more efficient
[11:39] Kayaker Magic: I hear there are recent improvements, and I will go check it out!
[11:39] Andrew Hellershanks: There are two test regions in a grid I help run that I will update to the latest ubODE.
[11:40] Misterblue Waves: what is different with ubODE than BulletSim that some things might not work if someone converted?
[11:40] Misterblue Waves: ubit and I tried to make them similar
[11:40] Misterblue Waves is enjoying the crowd today!!
[11:40] Andrew Hellershanks: nebadon, Misterblue: compatibility is the bit issue with changing physics engines.
[11:41] Andrew Hellershanks: This is the biggest crowd we have had for a meeting in ages.
[11:41] Andrew Hellershanks: 15 people here other than myself.
[11:42] Nebadon Izumi: yea nice to see the gears turning again :)
[11:42] Nebadon Izumi: I have been so crazy busy with Encitra lately
[11:43] Andrew Hellershanks: I've been working on updates to the addon OpenSimSearch and OpenSimProfile modules in the last few days.
[11:43] Andrew Hellershanks: They have been updated to work with PHP 7.
[11:43] Gavin.Hird links?
[11:43] paela argus: can be work too on php 5.3 ?
[11:44] Andrew Hellershanks: and
[11:44] Gavin.Hird thanks!
[11:44] Andrew Hellershanks: paela, yes they should work on PHP 5.3 but I haven't tested with that version. My computer runs PHP 7 now.
[11:44] Andrew Hellershanks: The PHP portion of the modules is now using PDO calls instead of the old mysql calls that were removed as of PHP 7
[11:45] paela argus: can test later and send you if work or not :)
[11:45] Andrew Hellershanks: ok. One person said there is a problem with a couple of features in the profile module but those features work fine in my Standalone.
[11:46] Marcus Llewellyn: Gavin, I'm looking into Mantis 8147. Etcd is a key-value store service, and I this module seems to provide an interface for it. It has code for configs in the INI, but I can't find a reference to it in either OpenSim.ini or OpenSimDefaults.ini.
[11:47] Andrew Hellershanks: The Show In Search button in the profile dialog box doesn't work but that doesn't have any impact on anything in world. It currently isn't worth the effort to make that button work. It would only affect profiles being accessed from a website.
[11:47] Mike.Lorrey I am trying to integrate a gifting function into a vendor and running of course into the problem of getting the key of an avatar who is not present, and not a resident of the grid the vendor is on. I would like to see a change in osGetAvatarKey to include a parameter that specifies the grid to search for a given avatars key, that would have the function do an http request to the specified grid.
[11:48] Gavin.Hird It would be great if new modules are better described than a short commit message when added
[11:48] Marcus Llewellyn: Yeah. There's one terse comment in the source for that module.
[11:49] Mike.Lorrey Agreed, Gavin, the amount of documentation anywhere on things, not just in the source, but in the wiki, is terribly sparse, makes it pretty unusuable.
[11:49] Arielle Popstar: potential privacy issues i would think on that Mike
[11:49] Mike.Lorrey Arielle, there are other existing functions that have way bigger privacy issues that the devs had no problem implementing
[11:49] Mike.Lorrey esp given that everyones key is right on their profile.
[11:50] Andrew Hellershanks: I suspect that most of the avatar key related script functions only work properly in the local grid. The name to key function would probably fail when passed the UUID of an avatar from another grid.
[11:50] Arielle Popstar: yes but allowing other grid access to that
[11:50] Marcus Llewellyn: When this module doesn't find a [Etcd] section in the configs, it should just return a null from it's Initialise function. So not sure why it's screwing up your standalone.
[11:51] Gavin.Hird Melanie, who added it should test it also in standalone, if not done
[11:51] Marcus Llewellyn: Would be easier to diagnose if you ran it in debug mode.
[11:51] Andrew Hellershanks: The Etcd module defaults to disabled state unless it gets the initialization things it needs.
[11:51] Marcus Llewellyn: YEah
[11:51] Mike.Lorrey I am able to send objects to people across the hypergrid if I have their key, no matter what grid they are from. The problem is getting their key.
[11:52] Marcus Llewellyn: So unless you've somehow enabled it accidentally, it shouldn't die like that.
[11:52] Arielle Popstar: i mean gifties from faraway places is always nice but still unless there is an ability to pre approve allowing that key out of grid
[11:52] paela argus: making less talkative ubode at startup simulation stirred a more even debug ..
[11:52] Mike.Lorrey and putting some nebulous concept of privacy ahead of freedom of movement, communication, and commerce in a platform that calls itself "open" makes it a sad joke.
[11:52] Gavin.Hird The standalone was master raw off the repository, compiled and started
[11:53] Arielle Popstar: when you send an object to someone it is pre approved/requested
[11:53] Andrew Hellershanks: Gavin, you are saying the Etcd module is "new"?
[11:53] Gavin.Hird yes
[11:53] Gavin.Hird
[11:53] Gavin.Hird is the commit it was added
[11:54] Mike.Lorrey For instance, come to find out that the http and email modules in the OS default package are defaulted to off or uninstalled is also problematic. It makes all sorts of usability problems.
[11:54] Andrew Hellershanks: oh, it is. Added May 31 according to the changelog entry
[11:54] Andrew Hellershanks: Mike, part of setting up a Standalone or grid enabled version of OS is going through all the settings.
[11:55] Gavin.Hird EtcdMonitoringModule.cs was added in this commit
[11:55] Gavin.Hird Committer date 2017-03-31 13:38:41
[11:55] Marcus Llewellyn: It would be interesting to see what the error looked like when run in debug mode, Gavin. That gives much more specific information on where to look in the code.
[11:56] Gavin.Hird I'll check it out
[11:56] Marcus Llewellyn: With this error, I can see that the module is a problem, but not why.
[11:56] Mike.Lorrey Which it appears that few grid operators actually care to go through, or even if they do, there is insufficient documentation in the source for them to have any clue of how to properly set those parameters up. I just got done helping one major grid fix their http module to refer to their own grid domain by their actual domain name instead of "ubuntu"
[11:57] Andrew Hellershanks: I'm not sure that the changing config part of Etcd will work when it comes to modules. Most don't check for, or expect, any changes after they were started.
[11:58] Andrew Hellershanks smiles at Mike.
[11:58] Andrew Hellershanks: I've gone through the INI files many times. Often when there has been a big update to the grid code.
[11:58] Arielle Popstar: it likely would be in Diva's hands to sort out those issues across grids
[11:58] Mike.Lorrey great, how can I contact Diva?
[11:58] Arielle Popstar: irc
[11:59] Marcus Llewellyn: Going through your INIs is pretty much just required before any upgrade. That's what diff is for. :)
[11:59] Mike.Lorrey seriously? Why not chipping messages out of stone while we're at it?
[11:59] Arielle Popstar: lolz
[11:59] Mike.Lorrey Yes Marcus, but if the ini isn't well documented, thats just a bs excuse
[11:59] Andrew Hellershanks: Marcus, yup. That's why I now keep my INI files as close to the original example files as possible to make it easier to do diffs and catch any rqeuired updates.
[11:59] Arielle Popstar: too bad they cant make a viewer out of irc that can access sims :)
[11:59] Kayaker Magic: LOL, come on Mike, IRC is like using a CB radio, not quite as bad as stone tablets.
[12:00] Marcus Llewellyn: If we're talking about core OpenSim's configs, it's riddled with comments. Those things are practically 75% documentation.
[12:00] Andrew Hellershanks: I've seen a couple of instances where people have established a bridge between in world messaging and an IRC channel.
[12:01] Arielle Popstar: Phoenix used to include it
[12:01] Marcus Llewellyn: It used to be fashionable for TPVs to put an IRC client right into the viewer. Waaay back in the days.
[12:01] Arielle Popstar: like 3 years ago
[12:01] Andrew Hellershanks: Marcus, yes, the INI files are mostly well documented. I still don't 100% trust that they show the proper default values for all the settings. I suspect one or two may not be showing the proper default value.
[12:02] Marcus Llewellyn: I agree that they may not get all of the update love they could.
[12:02] Isis.Ophelia (Mike she has answered me when I wrote her a message in Twitter, maybe try?)
[12:02] Mike.Lorrey given how much scripting of various sorts, esp to interact with websites, is essential to OS usage by everyday users, having http functions default to "off" is stupid.
[12:02] Mike.Lorrey Great, Isis, whats her twitter handle?
[12:02] Isis.Ophelia hmm @divacanto
[12:03] Mike.Lorrey thanks
[12:03] Andrew Hellershanks: Mike, which HTTP functions?
[12:03] Isis.Ophelia yw but remember you only got 140 characters haha
[12:03] Mike.Lorrey http query, http-in?
[12:03] Arielle Popstar: she may suggest you go through pipermail or very least nabble
[12:04] Mike.Lorrey I can tell you right now that THIS regions http parameters are not properly set up.
[12:04] Mike.Lorrey so, who is the grid expert who set this region up?
[12:05] Arielle Popstar: Mike, is this what you are talking about:
[12:05] paela argus: this region doesnt need request http i have desabled :p
[12:06] Andrew Hellershanks: Mike, I'm not aware of http query being disabled out of the box.
[12:06] Andrew Hellershanks: http-in can be more of a firewall issue.
[12:06] Marcus Llewellyn: I'm not sure I'm seeing a problem in the INI file, at least where llHttpRequest is concerned. The default I see is a *secure* one. Which is sensible. If someone wants to open that up, that's on them.
[12:07] Mike.Lorrey which by default makes many very useful scripts completely useless in such a grid.
[12:08] Kayaker Magic: llHTTPRequest works by default, but llRequestURL requires extra setup in the INI file that few people do in OSGrid.
[12:08] Marcus Llewellyn: I'm pretty sure you're gonna have an uphill battle convincing core devs that an insecure default is the best choice.
[12:08] Arielle Popstar: nods
[12:08] Mike.Lorrey Doesn't need to be insecure, but you can't call it open if its not open.
[12:09] Arielle Popstar: opensim is a platform that allows things to be opened
[12:09] Andrew Hellershanks: The main restriction to the use of llHTTPRequest is being able to call local services. This is to prevent people messing about with scripts in world that could pass data or requests to regions or Robust.
[12:09] Marcus Llewellyn: It is open. Just to private network ranges. Like a closed grid might use. Open grids are an anomaly, not the norm.
[12:09] Andrew Hellershanks: llHTTPRequest is enabled by default, AFAICR as long as you are accessing remote web servers.
[12:09] Mike.Lorrey There are far more hypergridded grids than closed grids
[12:10] Mike.Lorrey and no, its not working for accessing remote servers
[12:10] Marcus Llewellyn: Hypergrided !== open in the sense that grid services are exposed to all and sundry, like they are here on OSgrid.
[12:10] Andrew Hellershanks: I haven't found a grid where I couldn't use it.
[12:10] Kayaker Magic: Mike, are you having issues with llRequestURL or llHTTPRequest?
[12:10] Mike.Lorrey I have, including this one. Wright Plaza is also faulty
[12:11] Marcus Llewellyn: It's not faulty. It's secured.
[12:11] Mike.Lorrey Kayaker, thats one of the things I've been investigating on my hypergridding this week. It happens with both all over the hypergrid
[12:11] paela argus: its prohibited on all plaza in osgrid Mike
[12:11] paela argus: not a error
[12:11] Gavin.Hird how do you test it?
[12:11] Kayaker Magic: Like Andrew, I have rarely, perhaps never, found a grid where llHTTPRequest did not work.
[12:12] Andrew Hellershanks: I use the function in my vendors to record sales to an offworld database.
[12:12] Mike.Lorrey "Prohibiting" the proper use of any lsl function is by definition not about security but anal retentive paranoia
[12:13] Arielle Popstar: ~smiles~
[12:13] Andrew Hellershanks: You know the saying. Just because you are paranoid doesn't mean they aren't out to get you.
[12:13] paela argus: the http asks it to be disabled by default as notified and enabled if need be, many do not even save what its useful when it runs a region
[12:14] Andrew Hellershanks: The In function is disabled by default.
[12:14] Mike.Lorrey Again, levels of paranoia that would impress the North Korean government do not belong in any platform that pretends to call itself "open"
[12:14] Marcus Llewellyn: Since there's nothing stopping you from changing the config, I have a hard time seeing what the hubbub is.
[12:14] Marcus Llewellyn: Okay, hyperbole isn't gonna help your arguments. Heh.
[12:15] Mike.Lorrey Marcus I bet you are the person who invented the "threat level" bs in os functions that Kayaker loves so much
[12:15] Kayaker Magic: I assumed llRequestURL is disabled because it needs to know the URL to reach your system from the outside, something the INI file cannot guess and must be set up. Most OSGrid installers don't bother. That's not paranoia, just lazyness.
[12:15] Andrew Hellershanks: A number of the defaults are meant to keep people "safe" out of the box. If you know and understand the risks you can always change the settings.
[12:15] Marcus Llewellyn: You're clearly passionate about your opinion on that config. But I can pretty much promise you that the default will not change. You're gonna have to come to terms with that. That's just how it is. Sometimes life sucks.
[12:16] Sheera.Khan awwww, please keep it from getting personal attacks, will you?
[12:16] Misterblue Waves: there were problems in SL and early OS where people would use scripts to generate spam email and use scripts to spam sites
[12:16] Misterblue Waves: SL has the energy feature which slows down and/or delays some operations
[12:17] Misterblue Waves: those operations are not totally open because there were griefers who misused them
[12:17] Gavin.Hird They are about to only allow verified email addresses to use email functions
[12:17] Mike.Lorrey of course, thats not 'not open" that is regulated to limit abuse. regulation and prohibition are not the same thing
[12:17] Andrew Hellershanks: I just saw that message the other day
[12:17] Misterblue Waves: the question might be about how to constrain them so they can be used without them being misused
[12:18] Mike.Lorrey llHTTPRequest in this region is throwing a fail status code of 0 and failing to transmit data to my outside server.
[12:18] Andrew Hellershanks: There are many LSL and os functions that have built-in delays to limit the rate at which they can be used.
[12:18] Andrew Hellershanks: Some of the other rate limiting features mentioned in SL are not in place in OS.
[12:18] paela argus: yes the firewall not allow this here :)
[12:18] Misterblue Waves: turning the request off does prevent misuse ;-)
[12:19] Andrew Hellershanks: :)
[12:19] Mike.Lorrey yes and killing people prevents them from breaking the law, even if they never intended to
[12:19] Mike.Lorrey once again you people proved your utter uselessness.
[12:19] Marcus Llewellyn: It is up to people to run their grids/simulators as *they* see fit. As secure as they desire. The config allows for that choice. Forcing everyone into an insecure default is no answer.
[12:20] Misterblue Waves: the only alternative is turning it on for everyone all the time?
[12:21] Marcus Llewellyn: I"m not saying there aren't other solutions. I'm only stating that the code and configs, as they are in the present, are sane for what it is.
[12:21] Misterblue Waves: I thought we were talking about that the current resriction is too strong and it would be useful to be lesser
[12:22] Misterblue Waves: but Mike is, I think, suggesting a useful use case that the current configuration doesn't cover
[12:22] Kayaker Magic: There are (at least) 3 issues with HTTP out and in, and people often assume you were talking about one when you mean the other.
[12:22] Andrew Hellershanks: It is past the hour now. Are there any other topics for today or can we wrap it up?
[12:23] Andrew Hellershanks: yes, the outgoing httprequest always works, AFAIK. I haven't used the feature where you ask for a URL and I've only tested the In feature on rare occasions.
[12:25] Andrew Hellershanks: The defaults are set to keep users of OS reasonably safe for the most part. If the default on the HTTP functions is too restrictive for a given use case it is up to the person running OS to change it.
[12:25] Sheera.Khan As much as I like discussions, but throwing a tantrum if you don't immediately get what you want isn't going to help your point ...
[12:26] Marcus Llewellyn: +1
[12:26] Andrew Hellershanks: Sheera, right.
[12:26] paela argus: +3
[12:26] Arielle Popstar: wont go well with Diva especially
[12:26] Isis.Ophelia I understood Mike was talking about requests over the HG? he cant influence all owners to change for him. For I might have understood wrong :)
[12:26] Misterblue Waves: I'm open to suggestions about solution between on and off... but there would have to be constructions to prevent misuse while opening it up for useful use

Personal tools
About This Wiki