Chat log from the meeting on 2022-11-01

From OpenSimulator

Jump to: navigation, search
[11:00] Andrew Hellershanks: Hello, everyone.
[11:00] Ubit Umarov: hi
[11:01] Andrew Hellershanks: Hello, Arielle
[11:01] Motoko.Karu yes and no... The horns are still original.. I exported... modded... <or atleast tried>..then reimported as a new version...
[11:01] Ubit Umarov: again that time of year, hour changed here
[11:01] Ubit Umarov: usa will change next sunday?
[11:01] Motoko.Karu already? we change this coming weekend
[11:01] Andrew Hellershanks: I was going to point out that clocks change this weekend in North America, IIRC. Where have they already changed?
[11:02] Ubit Umarov: yeah usa always does it at arong time
[11:02] Ubit Umarov: :p
[11:02] Andrew Hellershanks: Only one more time change for the US (in the spring) and that will be the last time, IIRC.
[11:02] Ubit Umarov: we do it on last sundays of march and octuber, if i remember
[11:02] Arielle Popstar: oh that right/
[11:03] Ubit Umarov: usa will stop changing?
[11:03] Motoko.Karu we'll see... @andrew... I haven't heard any chatter about it, from our elected officials
[11:03] Arielle Popstar: what about ontario?
[11:03] Motoko.Karu there was a huge push.... earlier in the year.. if I remembre correctly.. then is fizzled out
[11:03] Andrew Hellershanks: Ah, ok. I thought Europe changed clocks two weeks after (or was it before) they changed in North America. I remember we usually have about two meetings affected by clock changes.
[11:04] Ubit Umarov: eu decided to stop change like 3 yrs ago...
[11:04] Ubit Umarov: then forgot about it :)
[11:04] Arielle Popstar: used to be 2 weeks before but this time just a week
[11:04] Andrew Hellershanks: Canada will probably do what they do in the US but I haven't heard anything about no more clock changes.
[11:05] Ubit Umarov: ofc if we used UTC on opensim grids, no issues with hour changes :)
[11:05] Arielle Popstar: Ford waiting for the state of New York he said
[11:05] Kayaker Magic hates daylight wastings time
[11:05] Andrew Hellershanks: I did hear that New York was part of the key to the stopping of clock changing.
[11:06] Ubit Umarov: some love it
[11:06] Ubit Umarov: it seems
[11:06] Andrew Hellershanks: I'm not a morning person so I like it when the sun sets later in the day.
[11:06] Arielle Popstar: only bothers me because it throws out the schedule between my partner and I
[11:06] Ubit Umarov: well think my own gov, decided to keep changing when eu was telling to sotp
[11:07] Ubit Umarov: well now my clock is at UTC + 0 :)
[11:07] Jamie.Jordan hi all
[11:07] Arielle Popstar: Hi Jamie
[11:08] Ubit Umarov: so..  what news do you have?
[11:08] Ubit Umarov: about opensim..
[11:08] Arielle Popstar: what did you fix this week?
[11:08] Ubit Umarov: a bug i made on ubode :)
[11:09] Ubit Umarov: spoted by ppl at vincent grid
[11:09] Arielle Popstar: broke and fixed the same day?
[11:09] Andrew Hellershanks: Hello, Jamie.
[11:09] Ubit Umarov: sadly no.. took many hours
[11:09] Ubit Umarov: they did a lot of test and did pinpont the exact commit
[11:09] Vincent.Sylvester More a "missed file" type deal I would say given it was just a minor change in the end, with big impact
[11:10] Ubit Umarov: yeha some typos are like that
[11:10] Ubit Umarov: but hard to spot
[11:10] Ubit Umarov: unless you sail crossing regions all day :)
[11:10] Vincent.Sylvester Basically wasn't removing the physics properly so after a while it clogged up the system as if there were a lot more physical objects there
[11:10] Arielle Popstar: lotsa lag
[11:10] Ubit Umarov: yeahh just a tiny little bit of physics part
[11:11] Ubit Umarov: things on collisions "map"
[11:11] Ubit Umarov: so it added more and more things to tests.. so lag...
[11:11] Ubit Umarov: well bugs...
[11:12] Vincent.Sylvester So far no more reported issues, so the fix seems to work
[11:12] Ubit Umarov: don't worry, i will break something soon
[11:12] Andrew Hellershanks grins
[11:12] Vincent.Sylvester Shows the importance of knowing the commit that caused it making it in the end just a few hours of debugging to figure out
[11:12] Arielle Popstar: keeps you busy
[11:13] Ubit Umarov: ahh fixed a old bug on LSL also
[11:13] Ubit Umarov: algo reported by vicent ppl  and the very next day here lol
[11:14] Ubit Umarov: seems a script did turn very popular
[11:14] Ubit Umarov: that one, not sure if mine.. it was there for ages
[11:14] Arielle Popstar: dont need Priscilla Kleenex with that sort of bug
[11:14] Vincent.Sylvester That was a fun one to figure out, lots of bad LSL code hiding a weird little bug of being a bit too liberal with types
[11:15] Ubit Umarov: yeah that line of code on the script was just bad..
[11:15] Ubit Umarov: but well did hit a bug we had, not doing proper cast of a velocity vector to lsl vector
[11:16] Ubit Umarov: the cpu and memory we waste bc that decision to use doubles on lsl  grrr
[11:17] Ubit Umarov: whatever
[11:17] Vincent.Sylvester Probably a good thing YEngine unearths this stuff in a more drastic manner though
[11:17] Vincent.Sylvester Maybe it'll mean less bad LSL code in future, one can dream
[11:17] Vincent.Sylvester Scripts some people write to work at all
[11:17] Ubit Umarov: also made more useless changes on dotnet branch
[11:18] Ubit Umarov: inc removing a bit more of XEngine that was still there
[11:18] Vincent.Sylvester One way making code run faster is removing parts of it, so it's progress heh
[11:18] Ubit Umarov: and a thing OSSLPrim that i totally forgot the use it had, if i ever did :)
[11:18] Andrew Hellershanks: :)
[11:19] Ubit Umarov: ( well that one i left in comment in case anyone remembers that needs it badly )
[11:20] Arielle Popstar: what breaks with removing xengine code?
[11:20] Ubit Umarov: for some reason OSSL was keeping a copy of several props of prims, encapsulated on that
[11:20] Ubit Umarov: possible for that other thing i did remove some yrs ago
[11:21] Ubit Umarov: Yengine sintaxe is more correct and less forgiving
[11:21] Andrew Hellershanks: OSSLPrim goes back to at least September of 2008.
[11:22] Arielle Popstar: more correct?
[11:22] Ubit Umarov: yes
[11:22] Arielle Popstar: do9es it matter?
[11:22] Ubit Umarov: more by ll rules
[11:22] Ubit Umarov: well even programming rules
[11:23] Ubit Umarov: see now Yengine does have break
[11:23] Ubit Umarov: actually do did XEngine LOL
[11:23] Arielle Popstar: ?
[11:23] Ubit Umarov: bug just bc XEngine did fail to detect it was a error
[11:24] Arielle Popstar: so it worked anyway in spite of bug?
[11:24] Ubit Umarov: and did jsut copy it to the intermediate c# code
[11:24] Ubit Umarov: so... kinda worked
[11:24] Arielle Popstar: fault tolerant?
[11:24] Ubit Umarov: just accident
[11:24] Arielle Popstar: more code should be fault tolerant
[11:24] Ubit Umarov: well someone was using it and was suprise Yengine did list it as its own new feature
[11:25] Ubit Umarov: it also did things like llDie;
[11:25] Ubit Umarov: forgot to check for the ()
[11:25] Vincent.Sylvester YEngine does tolerate some bad code design. Example you can use the same variable name globally and in local context and it does actually know which is which, at least until you start mixing them into each other
[11:26] Ubit Umarov: and ofc did fail to parse some math
[11:26] Ubit Umarov: like <3,4,5> - <a,b,c>
[11:27] Ubit Umarov: or similar ( there are several mantis on it )
[11:27] Andrew Hellershanks: Vincent, other programming languages let you do that. Not advisable unless one really knows what one is doing as it can be the cause of subtle bugs.
[11:27] Ubit Umarov: we had to put () there
[11:27] Ubit Umarov: vicent that is standard
[11:28] Vincent.Sylvester Yep, but XEngine will bark at you for using the same names regardless of context
[11:28] Ubit Umarov: and a reason why all tell to avoid globals on programs
[11:28] Ubit Umarov: and added namespaces, etc
[11:28] Vincent.Sylvester heh
[11:29] Vincent.Sylvester Just shows the different ways things parse, XEngine looking globally and YEngine being aware of context
[11:29] Ubit Umarov: if use same name on a local, then only that will be used, ignoring the global
[11:29] Ubit Umarov: standard thing
[11:29] Ubit Umarov: wil be used on that scope/method ofc
[11:29] Andrew Hellershanks: In some compiled languages you may get a warning about that type of use.
[11:30] Vincent.Sylvester YEngine being closer to LSL standard should improve quality of scripts, given you no longer have to work around the XEngine quirks
[11:30] Vincent.Sylvester Actually using sleep in a somewhat non-destructive manner is really nice
[11:30] Ubit Umarov: well and llSleep is safe :)
[11:30] Ubit Umarov: yeah
[11:31] Ubit Umarov: ofc not perfect..  but.. neither is vincent
[11:31] Vincent.Sylvester Birds of a kind flock together, must be some reason I attract all the bugs lately
[11:31] Ubit Umarov: yeah
[11:32] Ubit Umarov: but well 2 fixed
[11:32] Ubit Umarov: so i can make more
[11:33] Vincent.Sylvester I got some more stuff on mantis to re-test and see so may have some more minor changes depending on the results
[11:33] Vincent.Sylvester The whole double float thing has been tripping up minor rounding errors for years
[11:34] Ubit Umarov: yeah now and then ppl have surprise bc diferent round
[11:34] Ubit Umarov: but well what to say
[11:34] Ubit Umarov: it was even c# mentality
[11:34] Ubit Umarov: C3 as floats but the math lib only did double
[11:34] Ubit Umarov: c# has..
[11:35] Ubit Umarov: sure the intel FPUs where kinda double only
[11:35] Ubit Umarov: in fact a bit more than double
[11:35] Ubit Umarov: so they did all at same higher res
[11:36] Vincent.Sylvester It does make sense given the larger regions to have as much precision as possible though I would argue
[11:36] Ubit Umarov: but storage was difent saving 4 bytes per number
[11:37] Ubit Umarov: well but on region code all uses float
[11:37] Ubit Umarov: only LSL uses double
[11:37] Ubit Umarov: so.. bahh whatever
[11:38] Ubit Umarov: well the old FPU is now replaced on recent cpus my the SIMD math silicon
[11:38] Ubit Umarov: and on that, floats get own instructions
[11:38] Ubit Umarov: still most take the same time as double..  but a few are faster as float
[11:38] Ubit Umarov: so that no .net added a new math lib
[11:39] Ubit Umarov: MathF
[11:39] Ubit Umarov: ie math with floats
[11:39] Ubit Umarov: ( for now a lot is just calls to the old one :P )
[11:40] Ubit Umarov: well ages ago i changed our terrain storage from double to float
[11:40] Ubit Umarov: so on some blogs im considered a very bad programmer
[11:40] Ubit Umarov: ( i maybe... but not exactly bc of that :P )
[11:40] Vincent.Sylvester Time consuming to change, breaking existing scripts and losing precision on larger regions, kinda speaks heavily against change it now. I was thinking about adding some form of alternative math and just override the concerning LSL functions, but I don't think that is possible via modules
[11:41] Ubit Umarov: that was other spot where ppl decided to use doubles
[11:41] Ubit Umarov: to represent something that has like 16bits of precision on viewers
[11:41] Ubit Umarov: duhhh
[11:42] Ubit Umarov: and is sent with a loosy  jpeg like compressor
[11:42] Ubit Umarov: well details
[11:42] Vincent.Sylvester Worlds prettiest dumpster fire heh
[11:43] Ubit Umarov: well but our worse things are the XML based protocols
[11:44] Ubit Umarov: we cant' do anything about the ones to/from viewers
[11:44] Ubit Umarov: but ours are even worse
[11:45] Ubit Umarov: see a asset binary data ie sent from assets server to regions uuencode
[11:45] Ubit Umarov: ( ie base64 encoded)
[11:45] Ubit Umarov: that makes it 4/3 larger
[11:45] Ubit Umarov: plus all the time encodign and decoding..
[11:46] Ubit Umarov: so it fits on a ... ofc XML blabla
[11:46] Ubit Umarov: but again..  XML was the best thing for many back in first decade of 2000's
[11:46] Vincent.Sylvester Which internally isn't even as strict as it is supposed to be, so other parsers have trouble with it
[11:47] Vincent.Sylvester Search has been a pain to get right
[11:47] Ubit Umarov: ms also did love it
[11:47] Ubit Umarov: a bit better.. sure :)
[11:47] Vincent.Sylvester For some stuff it makes sense, but when you have more structure than actual data it becomes bloat
[11:48] Ubit Umarov: shh don't tell them that the 'j' on json means JAVA, that they hate...
[11:48] Andrew Hellershanks: Vincent, in what way has search been a pain?
[11:48] Vincent.Sylvester If you recall I redid most of it to figure out how it works and how to properly parse the data it gives
[11:49] Vincent.Sylvester There has been some nasty errors in parsing because of special chars and structure not being entirely correct
[11:49] Vincent.Sylvester I fixed most on the web end, but I been meaning to look over datasnapshot itself to find the structure issues
[11:50] Ubit Umarov: well protocols are a mess
[11:50] Ubit Umarov: dotnet offers several things to make microservices blabla
[11:51] Vincent.Sylvester Well the better threading that comes along with that might mean we can load regions more before they crash
[11:52] Ubit Umarov: some use protobuffers, other use  MessagePack, etc..
[11:53] Ubit Umarov: i just don't like any of those to start changing
[11:54] Vincent.Sylvester At least the documentation is being updated along with it for the most part, that used to be neglected so it was trial and error whenever a change was made
[11:54] Ubit Umarov: doc what? what is that?
[11:54] Ubit Umarov: ;)
[11:54] Vincent.Sylvester heh
[11:55] Vincent.Sylvester They still do the "this is deprecated now" without a word on what else to use or how one could work around it
[11:55] Vincent.Sylvester Just, it's gone, now f off lol
[11:55] Ubit Umarov: you will seen..  soon they will also do commits "cosmetics.."
[11:55] Ubit Umarov: will see
[11:56] Kayaker Magic: "Code in haste, document at leisure"
[11:56] Andrew Hellershanks: yup
[11:57] Ubit Umarov: what version is digi using mot?
[11:57] Ubit Umarov: well ofc own naming etc..
[11:57] Andrew Hellershanks: Another project says "this is deprecated. Use blah"
[11:58] Ubit Umarov: well i did hide a lot of such messages on dotnet6 branch
[11:58] Ubit Umarov: shhhh don't tell
[11:58] Ubit Umarov: like all the bitmap things.. bahh
[11:59] Ubit Umarov: ah hi did update the build instructions on that branch
[11:59] Ubit Umarov:
[12:01] Andrew Hellershanks: I noticed there had been changes to that file.
[12:02] Ubit Umarov: once can run opensim without mono installed now
[12:02] Ubit Umarov: ... run it on windows :p
[12:03] Ubit Umarov: kidding, on linux, installing dotnet and libgdiplus
[12:03] Vincent.Sylvester Still need to get the mess that is dotnet installed, plus the other dependencies usually shipped long with mono
[12:03] Ubit Umarov: only libgdi that debian 11 has as .deb
[12:04] Ubit Umarov: it seems
[12:04] Vincent.Sylvester Technically all just files so can fetch them from existing mono thing if need be
[12:05] Ubit Umarov: yeah it is there if mono is installed
[12:05] Vincent.Sylvester dotnet itself can be tricky to install I have found, dependency tree not setup well, but eventually you get it working
[12:05] Ubit Umarov: this boxes are on debian, installed easy
[12:06] Selby.Evans must fo -- bye all
[12:06] Ubit Umarov: cya
[12:06] Arielle Popstar: tc Selby
[12:06] Arielle Popstar: me too
[12:07] Ubit Umarov: well its time
[12:07] Ubit Umarov: 19:00 utc
[12:07] Ubit Umarov: 19:07
[12:08] Andrew Hellershanks: We are just past the hour. did someone else have any question/comment to make before we pack it in.
[12:09] Andrew Hellershanks: I don't see any signs that someone is typing so I will take that as a no.
[12:10] Andrew Hellershanks: Don't forget that clocks change in North America this coming weekend. If you forget you will be an hour early for next weeks meeting.
[12:10] Andrew Hellershanks: Thank you all for coming. See you next week.
Personal tools
About This Wiki