Chat log from the meeting on 2021-09-07

From OpenSimulator

Jump to: navigation, search
[11:03] Gavin.Hird @grid.xmir.org:8002: Hi Andrew
[11:03] Selby.Evans @grid.kitely.com:8002: Google search on Metaverse:  About 9,320,000 results
[11:03] Selby.Evans @grid.kitely.com:8002: Hi Andrew
[11:04] Kayaker Magic: Welcome Wendie!
[11:04] Gavin.Hird @grid.xmir.org:8002: Hi Wendie
[11:04] Wendie.Blackthorn @kinkyhaven.com:8002 nudges Kayaker
[11:05] Wendie.Blackthorn @kinkyhaven.com:8002: ♄ї ε√εяƴ♭◎∂ƴ  。◕‿◕。
[11:05] Selby.Evans @grid.kitely.com:8002: Hi Wendie
[11:05] Andrew Hellershanks: Hello, everyone
[11:05] Wendie.Blackthorn @kinkyhaven.com:8002: lots of names not loaded yet
[11:06] Gavin.Hird @grid.xmir.org:8002: all naes loaded here
[11:06] Gavin.Hird @grid.xmir.org:8002: names*
[11:06] Kayaker Magic: Hmm, I see all the names. Could it be 'cause you server is in Europe?
[11:06] Gavin.Hird @grid.xmir.org:8002: My server is in Europe
[11:07] Gavin.Hird @grid.xmir.org:8002: on the fringes of Europe even
[11:07] Ubit Umarov: what names avatar ones?
[11:07] Ubit Umarov: if so, nothing to do with servers location
[11:08] Wendie.Blackthorn @kinkyhaven.com:8002: names are loaded in chat but not on the name tags
[11:08] Kayaker Magic: Wendie's PC is in the USA, grid server in Europe, so double the transit time.
[11:08] Ubit Umarov: just old issue od viewers
[11:08] Ubit Umarov: as i said several times
[11:08] Ubit Umarov: on many ocasions viewers ask things to the worng region on teleports
[11:09] Wendie.Blackthorn @kinkyhaven.com:8002: the name tags look like (???).(???)
[11:09] Ubit Umarov: ie during a some period they ask abotu names, for example, the the tp start region
[11:09] Ubit Umarov: anywhere this means npcs names will just fail, because only their region knows them
[11:10] Kayaker Magic: If you TP to someplace nearby and back might that fix it?
[11:10] Gavin.Hird @grid.xmir.org:8002: what viewer do you use?
[11:10] Ubit Umarov: on HG tps all names will just fail
[11:10] Wendie.Blackthorn @kinkyhaven.com:8002: fs 6.4.21
[11:10] Gavin.Hird @grid.xmir.org:8002: ok
[11:10] Ubit Umarov: but when regions give no answer, viewers ask again MINUTES later
[11:11] Ubit Umarov: and that time to the correct region..
[11:11] Wendie.Blackthorn @kinkyhaven.com:8002: I see the names in chat so I'm not to worried about it
[11:12] Ubit Umarov: on same places they show (loading)
[11:12] Gavin.Hird @grid.xmir.org:8002: I set tags to only show briefly. I find it rather annoying to have the permanently on
[11:12] Ubit Umarov: older versions of opensim did return names like ummblablabla
[11:12] Ubit Umarov: that was bad, caused the viewer to actually cache that name for a long time :)
[11:13] Ubit Umarov: not anymore..
[11:13] Andrew Hellershanks: I remember those days. Haven't seen that in quite a while.
[11:13] Ubit Umarov: just can take a few minutes to see the real name
[11:13] Ubit Umarov: i removed any answer on 0.9.x
[11:13] Ubit Umarov: so viewers do ask again
[11:14] Ubit Umarov: ofc this is a bad issue
[11:14] Ubit Umarov: also on textures etc
[11:14] Ubit Umarov: dynamic textures do fail a lot
[11:14] Ubit Umarov: and those viewers do not ask for again :(
[11:14] Gavin.Hird @grid.xmir.org:8002: gosh
[11:15] Gavin.Hird @grid.xmir.org:8002: those pesky viewers
[11:15] Ubit Umarov: this is a issue since ever
[11:15] Ubit Umarov: no fix, because all works at SL
[11:15] Ubit Umarov: all textures at SL are grid wide, names etc
[11:15] Ubit Umarov: so does not matter where the viewer asks
[11:16] Ubit Umarov: so even the viewer devs that know the issue, have it on a very very low priority :)
[11:16] Ubit Umarov: but it is a deep fail on the teleports logic
[11:17] Ubit Umarov: somewhere
[11:17] Gavin.Hird @grid.xmir.org:8002: what logic?
[11:17] Ubit Umarov: exactly
[11:17] Ubit Umarov: :)
[11:18] Ubit Umarov: so what news do you have about opensim?
[11:19] Gavin.Hird @grid.xmir.org:8002: not much to report here - compiling viewer code as we speak
[11:20] Kayaker Magic: Gavin, how much work do you think it would be to clean up the dev environment of FS to compile with new dev tools?
[11:20] Ubit Umarov: ( gavin is daytunr not fs :p )
[11:20] Gavin.Hird @grid.xmir.org:8002: which new tools did you have in mind?
[11:21] Kayaker Magic: I've heard the dev environment is a bunch of old tools, difficult to set up a system to build it.
[11:21] Ubit Umarov: think tool ll several months to change to vs2017 i think
[11:21] Gavin.Hird @grid.xmir.org:8002: Win builds are on VS2017 and they spent about 1-2 years moving from 2013
[11:22] Gavin.Hird @grid.xmir.org:8002: loads of code that needed to be upgraded
[11:22] Kayaker Magic: Why is it so difficult?
[11:22] Gavin.Hird @grid.xmir.org:8002: first of all the vidwer code overall is old, old, old
[11:22] Andrew Hellershanks: 1-2 years to go from 2013 to 2017? That seems an incredibly long time.
[11:22] Gavin.Hird @grid.xmir.org:8002: then you have all the libs
[11:23] Ubit Umarov: ( i fixed this parcel stream )
[11:23] Gavin.Hird @grid.xmir.org:8002: it is a,long time because they found the coprocedures they had worked so much on did not compile at all in V2017 with the LL twist on boost
[11:23] Ubit Umarov: well there is another thing, it is not ms to decides when one shoud change compilers
[11:24] Gavin.Hird @grid.xmir.org:8002: so they first had to fix that with a different version of boost to make all that logic work
[11:24] Ubit Umarov: that is usually a very complex thing, no serius companie or dev just rushes that
[11:24] Gavin.Hird @grid.xmir.org:8002: then came all the other changes that MS had doen in VS2015 which was never fixed, and so on
[11:24] Ubit Umarov: smae with the .net framework nonsense
[11:24] Ubit Umarov: .net 5,0 ms now wants all to use it a total useless crap for openims
[11:24] Ubit Umarov: opensim
[11:25] Gavin.Hird @grid.xmir.org:8002: for theMac version is is basically a 10 manyear rewrite
[11:25] Kayaker Magic: what is 'boost'
[11:25] Gavin.Hird @grid.xmir.org:8002: it is a big library
[11:25] Ubit Umarov: for example because ms devs decided somethings .net framework had, is no longer needed
[11:25] Gavin.Hird @grid.xmir.org:8002: Linux version - unknown
[11:26] Ubit Umarov: and mono is a dead project
[11:26] Gavin.Hird @grid.xmir.org:8002: yes it is
[11:26] Ubit Umarov: that is as dead as .net framework
[11:26] Ubit Umarov: because ms said so
[11:26] Gavin.Hird @grid.xmir.org:8002: yep
[11:27] Ubit Umarov: ms wants all to migrate to .net 5  and excluded mono from it
[11:27] Kayaker Magic: But they are moving mono to something closer to .net
[11:27] Gavin.Hird @grid.xmir.org:8002: not an update for months on end
[11:27] Ubit Umarov: no they are not
[11:27] Ubit Umarov: they used mono to make their own thing
[11:28] Ubit Umarov: since ms declared .net framework 4.8 dead, so is mono basicly
[11:28] Ubit Umarov: in maintnance mode
[11:28] Ubit Umarov: again.. acording to ms
[11:28] Ubit Umarov: it is not that clear who really wons mono
[11:28] Kayaker Magic: So they are forcing Linux devs to switch from mono to a new thing that is .net 5.0?
[11:28] Ubit Umarov: yes they are
[11:28] Ubit Umarov: oyu need to install .net5 things
[11:29] Ubit Umarov: that will do less than mono does
[11:29] Ubit Umarov: lol
[11:29] Ubit Umarov: well for now
[11:30] Ubit Umarov: and you should had seen .net core 1.0 devs
[11:30] Ubit Umarov: those had killed a lot of fundamental things telling all apps would need to be rewritten
[11:30] Ubit Umarov: .net5 did add back a lot
[11:30] Ubit Umarov: just not all opensim for example needs
[11:31] Ubit Umarov: xengine is dead on that for example
[11:31] Ubit Umarov: so are dynamic textures, map generation.. etc
[11:32] Ubit Umarov: unless rewrittne using non ms tools ( same payed)
[11:32] Ubit Umarov: well..  this is abotu opensim and .net5 and why i just do not care about it
[11:33] Gavin.Hird @grid.xmir.org:8002: then we have Win 11...
[11:33] Ubit Umarov: .net framework should still have a few years of support
[11:33] Ubit Umarov: yeah win11 crap
[11:33] Ubit Umarov: i refuse to buy the stupid tpm chip
[11:33] Ubit Umarov: to have ms own my machine even more
[11:34] Kayaker Magic: I have a question about terrain loading. There is an option to load terrain near the avatar first, but it doesn't work the way I remember it used to.
[11:34] Andrew Hellershanks: A lot of people won't be able to run Win 11 with their current machines.
[11:34] Gavin.Hird @grid.xmir.org:8002: Ask not what Win 11 can do for you, ask what you can buy for Win 11.
[11:34] Kayaker Magic: It used to load a square around me, then increasing circles away from me.
[11:34] Ubit Umarov: ahh terrain load tools also dead on .net5 :p
[11:34] Gavin.Hird @grid.xmir.org:8002: I don't onw any machines that can run it
[11:35] Ubit Umarov: you mean to send near avatar on login?
[11:35] Kayaker Magic: Now it seems to just start the lawnmower mode near me, run north, then fill in the south side of me later.
[11:35] Kayaker Magic: Yes, on login or TP in.
[11:35] Ubit Umarov: yeah option is there, but a ugly disaster
[11:35] Ubit Umarov: forget it
[11:35] Ubit Umarov: viwers always draw the terrain
[11:35] Kayaker Magic: Forget? Meaning it will never be fixed?
[11:35] Ubit Umarov: no fog of war nothing
[11:36] Ubit Umarov: so you see even ugly gaps as you move around a large region
[11:36] Gavin.Hird @grid.xmir.org:8002: I never see those any more
[11:36] Ubit Umarov: cannot be fixed without some kind of fog of war or something
[11:37] Ubit Umarov: the option is there somewhere
[11:37] Ubit Umarov: i just disabled it by default
[11:37] Ubit Umarov: in fact terrain is sent close to the avatar, them the full sim
[11:37] Ubit Umarov: hmm or there is a option for that
[11:37] Ubit Umarov: but even another issue
[11:38] Ubit Umarov: viewers do not like to render the terrain, not having it lal
[11:38] Ubit Umarov: all
[11:38] Ubit Umarov: htye need to blend the textures etc etc
[11:38] Ubit Umarov: so for viewers terrain is a simple "prim" basicly
[11:39] Ubit Umarov: viwers also want to do tesselation on it etc etc
[11:40] Ubit Umarov: so not that happy with parcial data on it
[11:40] Gavin.Hird @grid.xmir.org:8002: if they didn't it ouwl look like crap
[11:40] Gavin.Hird @grid.xmir.org:8002: would*
[11:40] Ubit Umarov: yeah
[11:41] Ubit Umarov: so to end the answer.. the option is there on opensim.ini ( or defailts)  you can try it and see
[11:41] Ubit Umarov: ( defaults... )
[11:41] Ubit Umarov: oh you are awesome wendie ?
[11:41] Ubit Umarov: ;)
[11:42] Kayaker Magic: Names and titles must be working.
[11:42] Gavin.Hird @grid.xmir.org:8002: [InterestManagement] section in OpenSim.ini
[11:42] Ubit Umarov: i had no fail this time :)
[11:42] Gavin.Hird @grid.xmir.org:8002: BestAvatarResponsiveness is default
[11:42] Ubit Umarov: even see the lifeguard name
[11:42] Ubit Umarov: no gavin not that
[11:43] Ubit Umarov: that is other story
[11:43] Gavin.Hird @grid.xmir.org:8002: it is for the terrain
[11:43] Ubit Umarov: does not include terrain
[11:43] Ubit Umarov: no, it is for everything else lol
[11:43] Ubit Umarov: prims and avatars :)
[11:44] Gavin.Hird @grid.xmir.org:8002: I suggest you try the different settings as they significantly changes how the terrain loading happens too
[11:45] Gavin.Hird @grid.xmir.org:8002: But then again you removed some like Frontback
[11:45] Gavin.Hird @grid.xmir.org:8002: which wokred pretty great actually
[11:46] Ubit Umarov: SimpleAngularDistance should be a better option on UpdatePrioritizationScheme
[11:48] Ubit Umarov: think i did hide that terrain option very well :P
[11:49] Andrew Hellershanks: :)
[11:50] Ubit Umarov: SendTerrainUpdatesByViewDistance
[11:50] Ubit Umarov: section [Terrain]
[11:51] Kayaker Magic: That is the one I thought I had selected.
[11:51] Ubit Umarov: well it is ugly
[11:52] Andrew Hellershanks: It is the default setting
[11:52] Ubit Umarov: as i explained, viewers not made to have it
[11:52] Ubit Umarov: oops its?
[11:53] Gavin.Hird @grid.xmir.org:8002: I usually set draw distance to 160 meters rather than the default 128 as it helps loading neighbor regions
[11:53] Ubit Umarov: well not anymore
[11:54] Ubit Umarov: ( on a next commit )
[11:54] Ubit Umarov: shou duse 64m instead :p
[11:54] Gavin.Hird @grid.xmir.org:8002: no
[11:54] Andrew Hellershanks: hehe
[11:54] Ubit Umarov: ok 32m
[11:55] Andrew Hellershanks: I'm using 64 in my FS.
[11:55] Ubit Umarov: wlel abotu last week code changes
[11:55] Gavin.Hird @grid.xmir.org:8002: setting it to more than 128 seems to create a load sphere around the avatar larger than a standard region, touching the neighbors so they load
[11:55] Kayaker Magic: I'm set to 1024
[11:56] Ubit Umarov: i changed how the simple regions httpserver sends back things
[11:56] Ubit Umarov: in fact all http things to viewers
[11:57] Gavin.Hird @grid.xmir.org:8002: in the last changes you did?
[11:57] Ubit Umarov: i gave up of per client bandwidth control
[11:57] Ubit Umarov: just let the net do it
[11:57] Gavin.Hird @grid.xmir.org:8002: good
[11:57] Ubit Umarov: no not good :P
[11:58] Ubit Umarov: instead i feed the net in a interleaved way
[11:58] Ubit Umarov: it a bit of each thing, each time
[11:58] Ubit Umarov: so all get a bit, in turn
[11:59] Gavin.Hird @grid.xmir.org:8002: sounds like a great shcme for retransmision request
[12:00] Ubit Umarov: so its its a FIFO,  of responses,  take one, send up to 128KB, at end put it back if still not done
[12:00] Ubit Umarov: and keep rotating the fifo
[12:00] Andrew Hellershanks: Hello, Lorenz
[12:00] Lorenz VonMatterhorn: Hello everyone.
[12:01] Andrew Hellershanks: Hello, kanojo
[12:01] Gavin.Hird @grid.xmir.org:8002: Hi Lorenz
[12:01] Ubit Umarov: in fact the fifo has 3 fifos or queues
[12:01] kanojo k: hi everyone!
[12:01] Kayaker Magic: Welcome Kanojo, but you are a bit late!
[12:01] Ubit Umarov: requests are sent like up to 3 from one, 2 for other, and 1 from a 3rd
[12:01] Ubit Umarov: so there is some priority
[12:02] Ubit Umarov: textures and meshes are on that last one
[12:02] Ubit Umarov: becaus they are lowest priority things
[12:02] kanojo k: late for what?
[12:02] Ubit Umarov: like it or not :)
[12:02] Ubit Umarov: in future may need to slow things a bit more
[12:02] Ubit Umarov: to preserve cpu and net for the rest of simulator
[12:03] Ubit Umarov: also on this i see problems with http3.0
[12:03] Kayaker Magic: Kanojo: This is the OpenSim Developers Open House meeting.
[12:04] Ubit Umarov: as tou know clever google decided to use UDP on http
[12:04] Lorenz VonMatterhorn: I was right, all nerds here.
[12:04] kanojo k: lol
[12:04] Andrew Hellershanks: office hours. :)
[12:04] Kayaker Magic: LOL. Nerd and proud!
[12:04] Ubit Umarov: so http 3.0 uses UDP protocol at low level
[12:04] Ubit Umarov: this means that streams with be stopped by access to normal http pages
[12:05] Ubit Umarov: this did not happen because UDP as priority over TCP
[12:05] Ubit Umarov: clever google does not care... only whats their services to look fast
[12:06] Gavin.Hird @grid.xmir.org:8002: You mean make sure everything you do gets interrupted by their ads
[12:06] Ubit Umarov: this means out udp may also have negative impact..
[12:06] Ubit Umarov: yeah that also
[12:06] Gavin.Hird @grid.xmir.org:8002: not also
[12:06] Ubit Umarov: looking to SL this is fun
[12:06] Lorenz VonMatterhorn: Good works guys. And please continue developing.
[12:06] Gavin.Hird @grid.xmir.org:8002: to ensure they do
[12:07] Ubit Umarov: as you remember SL created a special udp protocol
[12:07] Ubit Umarov: lludp
[12:07] Ubit Umarov: original all comms bc viewers and servers where udp
[12:07] Ubit Umarov: then new lindens arrived telling that was bad and replaced by http
[12:07] Ubit Umarov: now google makes http be UDP
[12:07] Ubit Umarov: lol
[12:08] Andrew Hellershanks: :)
[12:08] Ubit Umarov: in fact in last years ll just made a crap moving things to http
[12:08] Ubit Umarov: instead of fixing and improving their llUDP
[12:09] Gavin.Hird @grid.xmir.org:8002: there is no UDP support on their CDN
[12:09] Ubit Umarov: google QUIC is there ( and other things) showing how udp can be used to make a reliable protocol
[12:09] Ubit Umarov: they falied to migrate even tp http2.0
[12:10] Ubit Umarov: well and now http2.0 is dead
[12:10] Gavin.Hird @grid.xmir.org:8002: no CGN support for http2
[12:10] Gavin.Hird @grid.xmir.org:8002: CDN*
[12:11] Ubit Umarov: well this last changes i made will need a revert if oiur http was http3.0
[12:11] Ubit Umarov: or we would had none stop rubber banding
[12:12] Ubit Umarov: but for now test and tell :)
[12:12] Gavin.Hird @grid.xmir.org:8002: I think you have bigger worries than http 3
[12:12] Ubit Umarov: rez should be a bit faster nwo
[12:12] Ubit Umarov: now
[12:12] Andrew Hellershanks: This hour went by very quickly today. Any last minute questions/comments before we wrap things up for today?
[12:13] Ubit Umarov: well those where the major code changes last week
[12:13] Andrew Hellershanks: Thank you for the update, Ubit.
[12:13] Ubit Umarov: ofc none of you did notice arriving here lol
[12:13] Andrew Hellershanks: :)
[12:15] Gavin.Hird @grid.xmir.org:8002: that is because our viewers use the technology PreRez - so we have already rezzed before we arrive
[12:15] Ubit Umarov: ahh
[12:15] Ubit Umarov: and decided to use a code schema  ms now calls "obsolete"
[12:16] Ubit Umarov: but their big star replacement schema is nothing more than wrapper to the "obsolete" code
[12:17] Ubit Umarov: allows cleaner source code, but does a very complex low level thing that uses even more resources
[12:17] Ubit Umarov: so, as i said, i wetn to the obsolete one
[12:17] Ubit Umarov: and clear increase of performance and mem used
[12:17] Ubit Umarov: like 0.1% improvements :p
[12:17] Gavin.Hird @grid.xmir.org:8002: sounds like my grandmothere's freezer. when she went we could find out what people ate in the 70s by digging down through the layers
[12:18] Ubit Umarov: worse
[12:18] Ubit Umarov: modern code is like that freeze actually
[12:18] Gavin.Hird @grid.xmir.org:8002: :-)
[12:18] Ubit Umarov: they add crap on to of crap
[12:18] Selby.Evans @grid.kitely.com:8002: bye all
[12:18] Ubit Umarov: and call it evolution
[12:18] Andrew Hellershanks: Bye, Selby.
[12:19] Ubit Umarov: cya selby.Evans
[12:19] Gavin.Hird @grid.xmir.org:8002: bye Selby
[12:19] Ubit Umarov: possible all netword code on linux is still based on the 1980's or 90's select or epoll
[12:20] Ubit Umarov: ye to find when sockets got data or are ready to send more
[12:20] Gavin.Hird @grid.xmir.org:8002: So I can still debug it on a PDp-8?
[12:20] Gavin.Hird @grid.xmir.org:8002: PDP
[12:20] Ubit Umarov: :p
[12:20] Ubit Umarov: :p
[12:21] Kayaker Magic: I have the parts of a PDP-11 in my barn. I'm going to mount the core memory on the wall.
[12:21] Ubit Umarov: well long long since i did bother look to linux kernel sources
[12:21] Andrew Hellershanks: Kayaker, you aren't going to get it running?
[12:21] Gavin.Hird @grid.xmir.org:8002: I bet that core would sell for a fortune
[12:21] Kayaker Magic: In my spare time.
[12:21] Ubit Umarov: late pdp 11 are small things
[12:22] Ubit Umarov: no core memory
[12:22] Kayaker Magic: Original PDP-11 is a CPU made out of Quad D flip-flop chips.
[12:22] Ubit Umarov: late was a single chip cpu :)
[12:22] Andrew Hellershanks: Kayaker, nice. :)
[12:22] Ubit Umarov: sure your talking about pdp 11 ?
[12:22] Gavin.Hird @grid.xmir.org:8002: The memory for IBM mainframes used in the US air traffic control system was organized as 1 MB of core per 19" rack. Each machine had 10 such racks.
[12:23] Andrew Hellershanks: We are straying off topic so it is time to wrap up the meeting for today.
[12:23] Andrew Hellershanks: Thank you all for coming. See you again next week.
Personal tools
General
About This Wiki