[11:02] Andrew Hellershanks: Hello, everyone.
[11:04] Lyr Lobo smiles in greeting
[11:05] Vincent.Sylvester @hg.zetaworlds.com:8002: I don't think we had commits this week or I'm blind
[11:07] Motoko.Karu @login.digiworldz.com:8002: lol
[11:07] Ubit Umarov: Oh yes... bad devs.. no commits
[11:07] Vincent.Sylvester @hg.zetaworlds.com:8002: lol
[11:08] Andrew Hellershanks: Someone must still be on "vacation". ;)
[11:09] Vincent.Sylvester @hg.zetaworlds.com:8002: Not much going on, probably have to go through mantis again and sort out some issues
[11:09] Vincent.Sylvester @hg.zetaworlds.com:8002: Think there is still some stuff there I wanted to look at
[11:09] Web Rain: Hi there :)
[11:09] Andrew Hellershanks: Hello, Web.
[11:09] Andrew Hellershanks: I've been rather side tracked with a couple of personal software projects of late.
[11:09] Vincent.Sylvester @hg.zetaworlds.com:8002: Same
[11:11] Lyr Lobo: Well there were fires and smoke
[11:11] Andrew Hellershanks: Does anyone have any questions, comments, or concerns?
[11:11] Lyr Lobo: hard to develop when working on breathing
[11:11] Andrew Hellershanks: Lyr, Are you near (enough) to some wild fire?
[11:12] Lyr Lobo: No, but the fires were in Portugal
[11:12] Lyr Lobo: I am often... had to evacuate once as I am in Colorado
[11:13] Ubit Umarov: yeah we had some of those... good we have a little rain now
[11:13] Lyr Lobo: but not right now
[11:13] Vincent.Sylvester @hg.zetaworlds.com:8002: In light of recent news I may offer some information. As you all should be familiar with by now there are interconnections between grids to fetch all manners of data. When they go away all that ends up going into the void. Took some time the other day to come up with a little log crawler to check and see how bad that actually has gotten over the years
[11:13] Andrew Hellershanks: It was a couple (or so) summers ago that there were wildfires in the northern part of my country more than 1,000 miles away but the smoke eventually made itself known to where Ilive.
[11:14] Vincent.Sylvester @hg.zetaworlds.com:8002: It's not as bad as I thought it would be, probably in part due to not much content going round, but there were a few bad ones in there still. You can crawl your region logs for NameResolutionFailure to see for yourself
[11:14] Ubit Umarov: joe how is movement inter/extrapolitaion going?
[11:15] Ubit Umarov: polation..
[11:15] Vincent.Sylvester @hg.zetaworlds.com:8002: Going through the actual data that causes those failures by way of external references is another matter. Crawling through assets and trying to find a way to adjust them is not fun
[11:16] Web Rain: i had this problem in my sim with some textures that i transfered from here to there... missing textures data grimlin even if the textures was there
[11:16] Vincent.Sylvester @hg.zetaworlds.com:8002: Alternatively one could presumably just change dns to spoof the dead endpoints and provide some sane data locally, but that has it's own implications
[11:18] Vincent.Sylvester @hg.zetaworlds.com:8002: If it gets to a point where all that http queue clogs up a region regularly there might be something internally that can be done, possibly through blacklisting or something
[11:18] Vincent.Sylvester @hg.zetaworlds.com:8002: I think httpclient already has some of that if a lot of requests fail, but it doesn't keep that between restarts I don't think
[11:18] Andrew Hellershanks: Vincent, what sort of data is getting lost? I'm guessing that are referring to when a grid shuts down.
[11:19] Vincent.Sylvester @hg.zetaworlds.com:8002: creator data mostly, but friends and groups are also affected negatively
[11:19] Ubit Umarov: hard to keep a automatic blacklist of dns fails etc
[11:19] Ubit Umarov: because dns does fail on valid things now and then
[11:19] Vincent.Sylvester @hg.zetaworlds.com:8002: Yeah
[11:20] Ubit Umarov: but myeb we could have manual maintained list
[11:20] Ubit Umarov: like for well known "defund" grids etc
[11:21] Ubit Umarov: ofc that will not prevent the hg standalones and similar short lived things
[11:21] Ubit Umarov: ( things == urls )
[11:22] Vincent.Sylvester @hg.zetaworlds.com:8002: Plus doing that for every region config, would be better as service something like mutelist, could steal some code from that heh
[11:22] Vincent.Sylvester @hg.zetaworlds.com:8002: Manual list in robust to distribute to regions
[11:23] Vincent.Sylvester @hg.zetaworlds.com:8002: That way you could make robust check again after a while if the endpoint is truly dead
[11:23] Vincent.Sylvester @hg.zetaworlds.com:8002: Have regions request the data once a day or so
[11:23] Vincent.Sylvester @hg.zetaworlds.com:8002: Kinda like dns heh
[11:23] Ubit Umarov: well at a point that list will be longer than assets :p
[11:24] Ubit Umarov: nahh dns does not do that
[11:24] Ubit Umarov: dns just expires things
[11:24] Vincent.Sylvester @hg.zetaworlds.com:8002: I mean as in TTL for blacklisted urls
[11:25] Vincent.Sylvester @hg.zetaworlds.com:8002: On the list I keep there are like 4000 grids with uptimes of less than 15%
[11:25] Ubit Umarov: we do have those
[11:25] Ubit Umarov: jsut not persisten
[11:25] Ubit Umarov: t..
[11:25] Vincent.Sylvester @hg.zetaworlds.com:8002: Judging from the logs the situation is not as bad as I feared it might be. Worst case was 15000 dns failures across 50+ regions
[11:26] Ubit Umarov: you do run a dns cache on yr grid right?
[11:26] Vincent.Sylvester @hg.zetaworlds.com:8002: Probably from months of log data, so nowhere near enough to clog up the http pipes
[11:27] Web Rain: and what if the dead urls was add to on fail [GatekeeperService] > AllowExcept ?
[11:28] Ubit Umarov: thats not for gatekeeper
[11:28] Vincent.Sylvester @hg.zetaworlds.com:8002: Regions don't see those, they just get a false on teleport requests there
[11:28] Vincent.Sylvester @hg.zetaworlds.com:8002: They don't get a "this is all banned btw" as return
[11:29] Web Rain: yep :(
[11:29] Ubit Umarov: yeah dead != banned also
[11:29] Kayaker Magic: I gota run to my next meeting! RL calls!
[11:29] Andrew Hellershanks: tc, Kayaker.
[11:29] Kayaker Magic: Buy all!
[11:30] Vincent.Sylvester @hg.zetaworlds.com:8002: At most I have seen one region request about 10 dead enpoints at once, but that's far enough from clogging pipes thankfully
[11:30] Web Rain: for local grid the assets are stored locally so... this problem is more an intergrid HG
[11:30] Vincent.Sylvester @hg.zetaworlds.com:8002: If you happen to store a ton of assets on a region from defunct ones though that might be a different story
[11:30] Ubit Umarov: fun his hat went first
[11:30] Ubit Umarov: :)
[11:31] Cuga.Rajal @rajal.org:9000: I can think of a couple grids that do that
[11:31] Cuga.Rajal @rajal.org:9000: but most don't
[11:32] Cuga.Rajal @rajal.org:9000: I've also seen grids where the grid owner "took ownership" in order to eliminate the DNS delay
[11:33] Ubit Umarov: or to "eliminate" other things :p
[11:33] Cuga.Rajal @rajal.org:9000: yep sokve one problem create another
[11:34] Web Rain: i have this message in the console?!? maybe is related... [WARNING]: EOC marker not found. Codestream is corrupted.
[11:34] Vincent.Sylvester @hg.zetaworlds.com:8002: Once it fails it is in limbo and it seems that there is no dummy data sent either, so that might be something that could be adjusted immediately. If the request fails, returning a kinda empty profile back to the viewer so that doesn't get stuck
[11:34] Vincent.Sylvester @hg.zetaworlds.com:8002: Web, that's from the maptile renderer, can be ignored
[11:34] Cuga.Rajal @rajal.org:9000: Vince that sounds like a good idea
[11:35] Web Rain: ha! ok :) is the missing textures problem that cause it
[11:35] orbert tatham: There are a number of places where timeouts could be helped with defaults of that sort
[11:36] Vincent.Sylvester @hg.zetaworlds.com:8002: Always have to check though where the data ends up cached and thus potentially be "wrong"
[11:36] Cuga.Rajal @rajal.org:9000: The 7.x viewer does seem to get stuck on a few things
[11:37] orbert tatham: Found a bug where the sea and sky get messed up when TPing around even within OSGrid
[11:38] Ubit Umarov: remember that the old issue of viewers requesting data to the wrong region on tps was never fixed
[11:38] Ubit Umarov: that may also impact enviroment
[11:39] Web Rain: agree, the env don't update after TP... i i jump to my sim i will keep the same env like here
[11:40] Ubit Umarov: possible it did update.. to the old region one :P
[11:40] Web Rain: and i have to log out/in to get the local env
[11:40] orbert tatham: Yeah, that sounds right
[11:40] orbert tatham: Didn't realize it was an old bug
[11:40] Ubit Umarov: old bad timing issues
[11:41] Ubit Umarov: same as failed avatar names
[11:41] Ubit Umarov: specially npcs
[11:42] Ubit Umarov: viewers get things from new region, but keep asking to the older one abotu those
[11:43] Ubit Umarov: that kills npc names, HG avatar names, local textures, HG textures.. etc
[11:43] Ubit Umarov: a issue there since "ever"
[11:44] Ubit Umarov: will never be fixed, bc those things are globla things at SL
[11:44] Ubit Umarov: global...
[11:44] Ubit Umarov: ( alos bc it is a timing related issue.. hard to debug )
[11:46] Ubit Umarov: ahh also a issue on nearby regions..
[11:46] Ubit Umarov: viewers may ask to main region about things on the child ones
[11:47] Ubit Umarov: so npc names, local textures, etc will fail
[11:47] Andrew Hellershanks: I received an email (from an IM) this past week that is sort of related to this topic. The person was asking if I if dotnet instead of mono is the culprit for all of the problems people are having in the hypergrid? The newer opensim versions are all dotnet now.
[11:48] Ubit Umarov: ppl always had problems with hg :p
[11:48] Andrew Hellershanks: yeah. More likely due to different versions of OS in source vs destination rather than dotnet vs mono.
[11:49] Ubit Umarov: and what "all the problems" are those..
[11:49] Cuga.Rajal @rajal.org:9000: When I HG TP across grids that are 0.9.3 or above it generally works fine for me
[11:50] Cuga.Rajal @rajal.org:9000: I always have issues with some regions in OSGrid that are running older versions
[11:50] Web Rain: It is a notable technological advance but not the miracle solution to all problems.
[11:50] Andrew Hellershanks: Looking at the clock I see we are about 10 minutes away from the top of the hour. If we can hold off other questions, etc for a moment please. Lyr has an announcement.
[11:50] Cuga.Rajal @rajal.org:9000: I have a couple too
[11:50] Andrew Hellershanks: Lyr, wait a moment for final comments coming in then you can go ahead.
[11:50] Andrew Hellershanks: Go ahead, Lyr.
[11:51] Lyr Lobo: Thank you, Andrew
[11:51] Lyr Lobo: I would like to say a few short words about a dear friend that we lost recently.
~
[11:51] Lyr Lobo: The Organizing Committee at the OpenSimulator Community Conference honors the memory of Olivetree Lighthouse, who passed away from the sudden onset of a brain tumor.
[11:51] Lyr Lobo: ~
She loved OpenSimulator and worked as a volunteer and moderator for the OSCC presenations and on the speaker booths in OSCC Expo Region 3.
[11:51] Lyr Lobo: ~
Olivetree, also known as PinkSamurai, will be missed.
[11:51] Lyr Lobo: We were close... worked together for a dozen years in virtual worlds...and played the World of Warcraft together with our conference teams at ISTE. Thank you
[11:52] Lyr Lobo: You may not have known her, but her mesh designs and signs are all over the place *smiles*
[11:52] Lyr Lobo: Thank you
[11:52] Cuga.Rajal @rajal.org:9000: Sorry to hear Lyr. *Hugs*
[11:52] Ubit Umarov: ( you can see Olivetree Lighthouse profile here at osgrid also )
[11:52] Motoko.Karu @login.digiworldz.com:8002: =(
[11:52] Andrew Hellershanks: I don't recall having met her. Sorry to hear of her passing, Lyr.
[11:53] Lyr Lobo: Thank you, Ubit *smiles* yes, she was in many grids and worlds...a gentle loving lady who loved the work that you do
[11:53] Web Rain: :( my condolences to his family and friends
[11:54] Lyr Lobo: her profile is great
[11:54] Lyr Lobo: Came to Virtual Worlds (SL then Opennsim) to Connect, Teach, Learn, Explore, and enjoy getting to know people and their creativity. I'm still in love with all of it and the fact that I get to do all this with people from around the globe.
[11:54] Lyr Lobo: end of my tribute to a dear friend
[11:54] Lyr Lobo: thank you
[11:54] Web Rain: her (sorry)
[11:54] Lyr Lobo: no worries, Web
[11:54] Lyr Lobo: *smiles*
[11:55] Lyr Lobo: she was on stage during the sessions at the conference
[11:55] Lyr Lobo: and http://conference.opensimulator.org/ call for proposals/
[11:56] Lyr Lobo: and there's the conference... if you have content to share or want to discuss community issues, please let me know and we'll schedule a session
[11:56] Lyr Lobo: back to you, Ubit and Andrew
[11:56] Web Rain: I think I know her but not personally :(
[11:56] Lyr Lobo: Yes *smiles*
[11:57] Andrew Hellershanks: ok, I probably have met or seen her. I'm not good with names. I have a tendency to forget them. Particularly so if it is someone I run across about once a year.
[11:57] Andrew Hellershanks: Thank you, Lyr.
[11:58] Andrew Hellershanks: Any other questions, comments, etc. for today?
[11:58] Cuga.Rajal @rajal.org:9000: I filed a bug report this week that I need to ask about. I also have a strange linkset that changes LI in a confusing way. that I want to ask about.
[11:58] Web Rain: yep :)
[11:58] Cuga.Rajal @rajal.org:9000: Been waiting a while here to ask
[11:58] Andrew Hellershanks: Floor is yours, Cuga.
[11:59] Cuga.Rajal @rajal.org:9000: \The bug report had to do with firing the object_+rez event after llRezAtRoot()
[11:59] Web Rain: just a good news to conclude :) 0 failure in a standalone HG under HTTPS at this moment
[11:59] Cuga.Rajal @rajal.org:9000: object_rez
[11:59] Web Rain: oupss sorry Cuga :)
[11:59] Joe Magarac: Next meeting,, gotta go....
[11:59] Cuga.Rajal @rajal.org:9000: I added sample code showing the issuehttp://opensimulator.org/mantis/view.php?id=9165
[12:00] Andrew Hellershanks: ok, Joe. See you again next week.
[12:00] Lyr Lobo: bye Joe
[12:00] Cuga.Rajal @rajal.org:9000: JeffzKelley chimed in and said that what I reported was expected behavior.
[12:00] Andrew Hellershanks: Cuga, the problem is when rezzing from a script or manually via build tools?
[12:00] Cuga.Rajal @rajal.org:9000: Can someone confirm?
[12:00] Cuga.Rajal @rajal.org:9000: rezzing from a script
[12:01] Cuga.Rajal @rajal.org:9000: during a state_entry() in one of the states
[12:01] Cuga.Rajal @rajal.org:9000: if you rez more than one thing, object_rez doesn't fire until after the last thing rezzes
[12:02] Cuga.Rajal @rajal.org:9000: instead of after each llRezAtRoot
[12:02] Cuga.Rajal @rajal.org:9000: is this expected behavior?
[12:02] Ubit Umarov: you are taliking abotu things happening on diferent scripts
[12:02] Cuga.Rajal @rajal.org:9000: If so I'll cancel bug report
[12:02] Cuga.Rajal @rajal.org:9000: no same script
[12:03] Ubit Umarov: and there is no timing relation btw on rez and the event
[12:03] Cuga.Rajal @rajal.org:9000: I added a delay between the two rezzing events, thinking there might be a timing issue, but I can add a whole minute and it doesn;t make a diff
[12:04] Cuga.Rajal @rajal.org:9000: object_rez is blocked from firing until state_entry() finishes
[12:04] Ubit Umarov: hmm and if i remember, on same script, state_entry() runs before on_rez
[12:04] Andrew Hellershanks: I would expect an object to maintain its link order regardless of how it is rezzed.
[12:04] Cuga.Rajal @rajal.org:9000: not on_rez
[12:04] Cuga.Rajal @rajal.org:9000: onject_rez
[12:04] Cuga.Rajal @rajal.org:9000: which fires when the script rezzes something
[12:04] Ubit Umarov: no
[12:05] Cuga.Rajal @rajal.org:9000: there was an error in my report, I meant object_rez
[12:06] Ubit Umarov: things run on threads..
[12:06] Ubit Umarov: some things run in same threads as main code, others are queued and will run when there is a free script thread
[12:06] Cuga.Rajal @rajal.org:9000: The workaround is to have the newly-rezzed object send back it's UUID to the object rezzing it to a listener
[12:07] Web Rain: just a stupid idea... add an llSleep(0.05) after each rez...
[12:07] Cuga.Rajal @rajal.org:9000: OK, I understand. object_rez is asynchronous with the llrezAtRoot methods.
[12:07] Ubit Umarov: others things like events entry and on_rez, are placed on a script events queue
[12:07] Cuga.Rajal @rajal.org:9000: Thanks for explaining, I'll cancel bug.
[12:07] Ubit Umarov: again will run when there is a thread..
[12:08] Cuga.Rajal @rajal.org:9000: Web I did that, I can add several seconds delay and same result
[12:08] Cuga.Rajal @rajal.org:9000: object_rez thread is blocked until state_entry finished
[12:08] Cuga.Rajal @rajal.org:9000: that is the issue
[12:09] Cuga.Rajal @rajal.org:9000: OK so I can;t rely on object_rez firing with any small delay after llrezAtRoot. That's what I needed to knw from this question. Thank you
[12:10] Cuga.Rajal @rajal.org:9000: Other issue is a strange linkset
[12:10] Ubit Umarov: no it is NOT blocked
[12:10] Cuga.Rajal @rajal.org:9000: I have a 1-LI mesh and link it to a 1-LI regular flat cylinder
[12:10] Cuga.Rajal @rajal.org:9000: and it becomes 30 LI
[12:10] Web Rain: i don't know but way work, write a rezzer function that rez the things and fire it at the end of the state_entry
[12:10] Cuga.Rajal @rajal.org:9000: Ubit my code example doe block it, can you check?
[12:10] Ubit Umarov: it wil wait for a free thread to do it...
[12:11] Cuga.Rajal @rajal.org:9000: I added a minute delay and it still didn;t fire, with no CPU load or anything
[12:11] Cuga.Rajal @rajal.org:9000: repeatable
[12:11] Ubit Umarov: ( if there is none. it may look like "blocked", since it will need that same thread to do it )
[12:12] Cuga.Rajal @rajal.org:9000: the underlying reason isn;t as important as the fact that it's expected, and can;t be relied on to fire until the method is finished.
[12:12] Ubit Umarov: ohh
[12:12] Cuga.Rajal @rajal.org:9000: Did I get that right?
[12:13] Ubit Umarov: in fact it even used a thread from main threads pool to do the rez
[12:13] Cuga.Rajal @rajal.org:9000: It fires immediately when state_entry finishes
[12:13] Ubit Umarov: ...
[12:13] Ubit Umarov: Util.FireAndForget(x =>
{
Quaternion wrot = rot;
wrot.Normalize();
[12:14] Cuga.Rajal @rajal.org:9000: I am just trying to get a yes or no answer
[12:14] Cuga.Rajal @rajal.org:9000: Is this a bug or not?
[12:14] Cuga.Rajal @rajal.org:9000: Expected or not?
[12:14] Ubit Umarov: you are asking about unclear timing relations
[12:15] Web Rain: i think the problem is the order of the execution and the events availability when the state_entry is runing...
[12:15] Ubit Umarov: and those are not at all well defined
[12:15] Cuga.Rajal @rajal.org:9000: As it saysd in the bug report, it doesn;t matter how long of a delay I put between the 2 rez operations. object_rez doesn;t fire until state_entry finishes, and then fires immediately
[12:15] Web Rain: state_entry is not the best place to fire the things
[12:15] Ubit Umarov: well each rez also has a predeined delay
[12:16] Cuga.Rajal @rajal.org:9000: Web it's the only good place when you use states
[12:16] Cuga.Rajal @rajal.org:9000: rez zstuff on state_entry, delete the stuff on state_exit
[12:16] Cuga.Rajal @rajal.org:9000: I do that, I'm just using a different method to get the rezzed objects UUIDs
[12:16] Cuga.Rajal @rajal.org:9000: as a workaround.
[12:17] Cuga.Rajal @rajal.org:9000: Back to the mesh dome and prim linkset.
[12:17] Ubit Umarov: ohh
[12:17] Ubit Umarov: and anohter thing you are missing
[12:17] Cuga.Rajal @rajal.org:9000: Is it normal to go to 30 when linked and back to 2 LI unlinked? I just haven't seen that before. Happens even if I set the mesh to Physics Type None
[12:18] Ubit Umarov: a script only has one event running at any given moment
[12:18] Cuga.Rajal @rajal.org:9000: Ubit I think that rings the bell
[12:18] Cuga.Rajal @rajal.org:9000: explains it
[12:18] Cuga.Rajal @rajal.org:9000: OK it has to finish that event first
[12:18] Lyr Lobo: great
[12:18] Ubit Umarov: so.. yes.. object_rez(key id) will only will run AFTER leaving entry..
[12:19] Cuga.Rajal @rajal.org:9000: llListen must buffer the incoming messages while other threads are running
[12:19] Cuga.Rajal @rajal.org:9000: OK thanks for explainning.. That m,akes a lot more sense now
[12:20] Cuga.Rajal @rajal.org:9000: The LI change 2 -> 30 from linking was something I thought only happened in SL but maybe I'm wring
[12:20] Lyr Lobo thanks everyone and tiptoes out
[12:20] Ubit Umarov: yeha scritps do not have internal multitask..
[12:20] Ubit Umarov: only one event at a time
[12:20] Andrew Hellershanks: Bye, Lyr. See you again next week.
[12:20] Cuga.Rajal @rajal.org:9000: Have you see that before in OS or is there a reference somewhere on how LI changes with linking
[12:20] Web Rain: @++ Lyr :)
[12:21] Cuga.Rajal @rajal.org:9000: take care Lyr
[12:21] Ubit Umarov: well LI in opensim is just indicative
[12:22] Ubit Umarov: and at sl it also has "strange" jumps
[12:22] Cuga.Rajal @rajal.org:9000: much stranger in SL
[12:22] Cuga.Rajal @rajal.org:9000: The nubmber made sense. It was a 30-object model in Blender and I joined them into one, optimized, uploaded with a Cube physics shape
[12:22] Ubit Umarov: prims without mesh have li of 1 mostly
[12:22] Ubit Umarov: then with mesh get value for the mesh..
[12:22] Cuga.Rajal @rajal.org:9000: But it was 1LI everywhere, until I linked it to a regular flat cylinder and made the mesh Physics None
[12:23] Ubit Umarov: but only if some other things are in action..
[12:23] Cuga.Rajal @rajal.org:9000: and the 2 prims became 30 LI
[12:23] Ubit Umarov: i had a little car with torus
[12:23] Cuga.Rajal @rajal.org:9000: no physics, no textures, glow, scripts
[12:23] Ubit Umarov: was fine.. less than 30 li... all things counted just as prims
[12:24] Ubit Umarov: i did set one prim to shape type none... all the torus where counted as mesh, not prims...
[12:24] Ubit Umarov: li went skyhigh.. the car to the trash
[12:24] Cuga.Rajal @rajal.org:9000: Ive seen a simple regular blow up to 100s of LI by going physical
[12:24] Web Rain: about linksets, the viewer shows wrong link numbers :(
[12:25] Ubit Umarov: terhe is no protocol to tell number of prims
[12:25] Cuga.Rajal @rajal.org:9000: Yeah you have to use a script to get the right numbers
[12:25] Ubit Umarov: viewers have to count the lludp packets
[12:25] Web Rain: but the links are corrects
[12:25] Ubit Umarov: and keep the count right with changes..
[12:26] Cuga.Rajal @rajal.org:9000: I think that ay have been fixed in a recent FS update, I have recently checked and it's been reportig correct
[12:26] Cuga.Rajal @rajal.org:9000: after years of not
[12:26] Ubit Umarov: same abotu link number
[12:26] Ubit Umarov: viewer will assign a number based on the order he got the lludp packet
[12:26] Ubit Umarov: that has not relation to the sending order
[12:27] Ubit Umarov: has no...
[12:27] Ubit Umarov: so those things wil only work on a very good internet weather
[12:27] Web Rain: no FS7 for me :( there is a way to disable PBR in it?
[12:27] Ubit Umarov: no pbr is integral part of 7
[12:28] Web Rain: i have the same under localhost so, i don't think the quality of the web is involved
[12:28] Cuga.Rajal @rajal.org:9000: i have PBR blocked on mine. It's still slower.
[12:28] Ubit Umarov: actual inside same machine udp can be lost
[12:28] Ubit Umarov: or lose order
[12:28] Cuga.Rajal @rajal.org:9000: You can effectively block PBR in preferences
[12:28] Ubit Umarov: no
[12:29] Cuga.Rajal @rajal.org:9000: no?
[12:29] Ubit Umarov: the rendering engine was changed for pbr
[12:29] Web Rain: ha cool! i have just a poor i5 i don't want to burn it hihihi
[12:29] Ubit Umarov: and that is what has the impact..
[12:29] Ubit Umarov: evne if you hide processing of some pbr
[12:29] Cuga.Rajal @rajal.org:9000: Right. You aren't disabling PBR, just supprerssing diaplaying it
[12:30] Ubit Umarov: yeah
[12:30] Cuga.Rajal @rajal.org:9000: That does help somewhat for a slower computer but still slower than 6.x
[12:30] Web Rain: there is a way (in theory) to replace UDP with TCP?
[12:30] Ubit Umarov: why ? tcp is slower
[12:31] Cuga.Rajal @rajal.org:9000: I'm guessing thats what they are doing with the mobile viewer
[12:31] Ubit Umarov: gooble did even replace tcp by udp on http/3
[12:31] Cuga.Rajal @rajal.org:9000: in SL
[12:31] Ubit Umarov: well not pure udp...
[12:31] Web Rain: yep but can be SSLed XD
[12:31] Ubit Umarov: like lludp is not pure udp
[12:32] Ubit Umarov: gooble? google :)
[12:32] Web Rain: blue boxes, script IMs...
[12:32] Cuga.Rajal @rajal.org:9000: SSL is an order of magnitude more data going through the wire for the same end result
[12:32] Cuga.Rajal @rajal.org:9000: So inly makes sense when data has sensitive info
[12:33] Web Rain: yep agree but a bit more data is cheap as price for the security
[12:33] Cuga.Rajal @rajal.org:9000: more data makes a platform inaccessible to more than it helps
[12:33] Ubit Umarov: not sure if ssl is that much more data, not couting the negociation
[12:33] Ubit Umarov: it is a lot more CPU usage..
[12:33] Ubit Umarov: in many cases data is just the same
[12:34] Ubit Umarov: depending on the used cypher
[12:34] Web Rain: yep there is +/- 125ms add to the things times
[12:34] Ubit Umarov: the old basic one was just XOR
[12:34] Cuga.Rajal @rajal.org:9000: data throughput is the limiting factor for many grid owners
[12:34] Ubit Umarov: wel negotiation adds more data and a few more roundtrips
[12:35] Ubit Umarov: also connections to other sites
[12:35] Ubit Umarov: to validate the cert chain
[12:35] Cuga.Rajal @rajal.org:9000: there are SSL-specific DoS issues
[12:36] Ubit Umarov: ofc webrtc enforces ssl bahh
[12:36] Web Rain: about this! is possible to directly use let's encrypt cert but there is a permissions problem because the cert are owned by root and must stay
[12:36] Ubit Umarov: damm religions
[12:37] Ubit Umarov: only some can use let's
[12:37] Ubit Umarov: you need to own the domain
[12:37] Cuga.Rajal @rajal.org:9000: pay for SSL certs
[12:37] Ubit Umarov: yeah and per year...
[12:37] Web Rain: yes i know ^^ is for this the selfsigned kung-fu hihihi
[12:38] Web Rain: but robust can run under valid certs
[12:38] Ubit Umarov: webrtc forbids selfsigned
[12:38] Ubit Umarov: ( and everything is changing to forbid it also )
[12:38] Web Rain: there is 5 or 6 free certs providers now
[12:39] Ubit Umarov: well discord moved away form ssl
[12:39] Ubit Umarov: they use other lighter encript schema
[12:39] Cuga.Rajal @rajal.org:9000: Discord's user agreement is has unsatisfactory privacy for a that kind of service
[12:40] Ubit Umarov: bc ssl on voice just does not scale
[12:40] Cuga.Rajal @rajal.org:9000: They make money selling user data, it's been documented in YT videos
[12:40] Web Rain: i used certbot that automatically create and renew the certs when is time to do
[12:40] Ubit Umarov: its just wasted CO2
[12:41] Cuga.Rajal @rajal.org:9000: hey I need to go folks.. Thank you for answering my questions
[12:41] Ubit Umarov: you know that if you do not pay $ for a service, you are paying with your data
[12:41] Cuga.Rajal @rajal.org:9000: Have a great week!
[12:41] Andrew Hellershanks: I was about to say it is time to wrap things up for today. We are well past the half hour past the hour mark.
[12:41] Andrew Hellershanks: Thank you all for coming. See you again next week.
