Chat log from the meeting on 2022-03-15

From OpenSimulator

Revision as of 10:44, 18 May 2022 by Tampa (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
[11:00] Kayaker Magic: OMG, you know everyone in Europe is going to be an hour late to this meeting, right?
[11:00] Selby.Evans hi everyone
[11:01] Andrew Hellershanks: I'll go poke Ubit.
[11:01] Kayaker Magic: Hi Selby!
[11:01] Michael.Christopher Hello!
[11:02] Andrew Hellershanks: Hello, everyone.
[11:02] Andrew Hellershanks: I'll wait a few minutes to see if we get some late arrivals today.
[11:03] Kayaker Magic: For the record, I'll repeat: Everyone in Europe is going to be an hour late for this meeting because of Daylight Wastings Time in California.
[11:03] Andrew Hellershanks: :)
[11:03] Michael.Christopher :))
[11:03] Andrew Hellershanks: I didn't remember it was this past weekend or I would have mentioned it at last weeks meeting.
[11:04] Kayaker Magic: It snuck up on me Sunday.
[11:06] Michael.Christopher i didn't even realize till someoen commented monday - all my timepieces autoupdate, and I work so late and get up so early I always get up tired.  I didn't even realized I'd lost an hour :))
[11:08] Ubit Umarov: hi
[11:08] Kayaker Magic: I have a clock that auto-updates, using the pre-2007 formula (my stupid government changed it in 2007 so we're 2 weeks off from Europe).
[11:08] Andrew Hellershanks: Hello, Ubit
[11:09] Ubit Umarov: us changed very soon
[11:09] Michael.Christopher hey ubit
[11:09] Kayaker Magic: So that clock doesn't change on time, so I change it manually, then two weeks later it changes again.
[11:09] Ubit Umarov: here will be on last sunday this month
[11:09] Ubit Umarov: ie next 27th
[11:09] Andrew Hellershanks: yes, the clocks move forward earlier in the year and backwards later in the year than in the not too distant past.
[11:10] Michael.Christopher that must be so fun kayaker :/
[11:10] Kayaker Magic: so for 2 or three weeks I have clocks that are 3 hours apart. Until I got a modern cellphone I had no idea what time it was during this period.
[11:10] Ubit Umarov: guess eu ppl doesnt know grid time change, as i didn't :)
[11:11] Ubit Umarov: saved by plugh :)
[11:11] Andrew Hellershanks: :)
[11:11] Ubit Umarov: why doesn't usa use eu rules on that? gezz
[11:11] Kayaker Magic: We used to!
[11:12] Ubit Umarov: well eu did plan to stop changing
[11:12] Andrew Hellershanks: We always have about a two week period between when the clocks change in NA vs when they change in EU.
[11:12] Kayaker Magic: Then congress changed it in 2007 because someone thought it would save energy to change it by 2 weeks.
[11:12] Andrew Hellershanks: There is talk of no longer changing in NA as well but they are still talking about it.
[11:12] Ubit Umarov: not sure this changes same any energy now
[11:13] Ubit Umarov: same? save!!
[11:13] Kayaker Magic: Yeah, the California government has been promising to stop changing. They have been talking for years, might get around to it soon.
[11:13] Kayaker Magic: Then my new clocks will be wrong....
[11:13] Ubit Umarov: there was some mess and eu just delayed decision for ever :)
[11:13] Andrew Hellershanks: Use UTC time and you never have to change the clocks. ;)
[11:14] Ubit Umarov: i did not like the fact countries could pick to stay on summer time or winter
[11:14] Ubit Umarov: that would make the time confusing all around
[11:14] Kayaker Magic: China has one time zone for the whole country. Beijing time.
[11:15] Ubit Umarov: well here whne we joined eu, gov decided to put us in berlin time..  central eu time
[11:15] Jagga Meredith: voted down in an Alberta Canada referendum
[11:15] Ubit Umarov: we have been like that for 2 years.. and almost had a revolution :p
[11:16] Ubit Umarov: so back to our more natural time..  ie west ie utc :)
[11:16] Ubit Umarov: ie bst ie gmt uff
[11:16] Ubit Umarov: so many names for same thing :)
[11:16] Kayaker Magic: zooloo time
[11:17] Ubit Umarov: well time zones are political things, so each has own rules
[11:17] Ubit Umarov: bst is uk, west is eu.. so..  well wahtever
[11:18] Jagga Meredith: I may need to leave early to take the kittycat into the shop for repairs
[11:18] Jagga Meredith: no issues
[11:18] Andrew Hellershanks: Jagga, I hope any needed repairs are minor.
[11:19] Kayaker Magic: I've been looking at fsassets and how assets are stored, and noticed that images are gz compressed, even though they are already compressed J2K data!
[11:19] Jagga Meredith: infected cut on her back
[11:19] Michael.Christopher oh ouch
[11:19] Kayaker Magic: The first few images I looked at got smaller when I gzdecoded them!
[11:19] Kayaker Magic: so I did a test on a small grid and generate the following statistics:
[11:19] Ubit Umarov: wel here we are at wet, will be at west :)
[11:19] Kayaker Magic: 95,082 images, 49.1% were larger after gzipping
860 were the same size
1,884,482 wasted bytes in J2K files that got larger when gzipped, 
223,436,677 saved bytes of storage 1.8% of the original J2K data
Total size of all J2K compressed images before gz compression was 12,394,591,143
22.296327590942 seconds were wasted gzdecoding all the images.
[11:20] Ubit Umarov: fsassets does fzip?
[11:20] Ubit Umarov: or gzip
[11:21] Ubit Umarov: well can't change that now on older grids
[11:21] Kayaker Magic: they are GZ encoded. gunzip can read them, but I am using PHP to do that and calculate the statistics.
[11:21] Ubit Umarov: but it is silly yes
[11:21] Andrew Hellershanks nods
[11:21] Ubit Umarov: mesh are also already gzipped
[11:22] Kayaker Magic: Yeah, but mesh have zipped and unziped sections. The few I looked at did benifit from gz compressiion.
[11:22] Kayaker Magic: I'm going to do the statistics on the mesh next.
[11:22] Ubit Umarov: unzipped is minor part
[11:23] Andrew Hellershanks: A mesh may compress but the associated textures may already be compressed.
[11:23] Ubit Umarov: well that was made thinging abtu text assets
[11:23] Ubit Umarov: like scripts and notecards i guess
[11:23] Kayaker Magic: Small text files always get larger when gzipped, because the file has a header!
[11:24] Ubit Umarov: no thye do not
[11:24] Ubit Umarov: unless single letter :P
[11:24] Ubit Umarov: ok very small
[11:24] Ubit Umarov: but on those hard to have ziiped or not
[11:27] Kayaker Magic: On a related subject: The mesh assets are stored as BINARY LLSD files. Does opensim read those, or just pass them on to the viewer?
[11:28] Ubit Umarov: they are parsed on uploads
[11:28] Ubit Umarov: but stored basicly has viewer sent
[11:28] Ubit Umarov: as
[11:28] Kayaker Magic: I wondered if there is any code in OpenSim that can decode those BINARY LLSD files. Is there a class for doing that? Something that converts them to XML? Reads them into an XML like DOM?
[11:29] Ubit Umarov: i already told they are parsed, that means decoded and checked
[11:29] Kayaker Magic: If Gavin were here, I'd ask him how the viewer does this...
[11:29] Ubit Umarov: no point on converting to xml
[11:30] Ubit Umarov: opensim has no use for that
[11:30] Michael.Christopher ive been on UI this week, so I was hoping to see gavin too lol
[11:30] Kayaker Magic: When are they parsed and checked? Not every time they are needed by the viewer? 'cause the viewer has to do it's own parsing.
[11:31] Ubit Umarov: mesh formats based on xml, like collada and gltr(?) are pure fails
[11:31] Ubit Umarov: they are just 2 large  to transmite
[11:32] Ubit Umarov: comparing to other binary formats
[11:32] Kayaker Magic: Well, LL documents LLSD format as an XML format, then in an appendix they say (oh, there is a binary version also, that we use everywhere)
[11:32] Ubit Umarov: well gl?? does have a terrible binary format
[11:32] Ubit Umarov: but meshs are not LLSD xml
[11:32] Ubit Umarov: there are 3 types of LLSD
[11:32] Kayaker Magic: correct, they are BINARY LLSD
[11:32] Ubit Umarov: xl, notation and binary
[11:33] Ubit Umarov: xlm..
[11:33] Andrew Hellershanks: Kayaker, Ubit already stated that mesh files are parsed and checked when uploaded.
[11:33] Ubit Umarov: yes we do parse to do minimal checks and estimate upload cost and LI
[11:34] Kayaker Magic: Yes, but when and why? Opensim could just send and recieve mesh a binay blobs to the viewer, from the viewer into the assets, etc.
[11:34] Ubit Umarov: later physics engines and decode the physics parts
[11:34] Ubit Umarov: warp3d map also decodes them, ofc
[11:34] Ubit Umarov: as i said, that MUS be done at upload
[11:35] Andrew Hellershanks: Kayaker, Parse and checked when uploading in to OpenSim. After that they would just be passed to the viewer as is.
[11:35] Kayaker Magic: OK. Is there a class or module or something to decode them?
[11:35] Ubit Umarov: you do not want us to accept any crap a viewr sends, do you?
[11:35] Ubit Umarov: meshcosts.cs
[11:35] Ubit Umarov: for example
[11:35] Ubit Umarov: warp3d uses primmesher
[11:35] Ubit Umarov: from libomv ???
[11:36] Ubit Umarov: physics use ubodemesmerizar or meshmerizer
[11:36] Andrew Hellershanks: It has been a while since I uploaded a mesh. I don't recall noticing a cost estimate prior to the final upload step.
[11:36] Ubit Umarov: yes ew know you still only use 0.8.2
[11:37] Ubit Umarov: ofc default upload base price is 0
[11:37] Ubit Umarov: well on last week code changes
[11:38] Ubit Umarov: VIncent did try to clean up some source files
[11:38] Ubit Umarov: still with non standard indentation.. tabs etc
[11:39] Ubit Umarov: i did some work on a new email module
[11:39] Ubit Umarov: decided to use a older version of mailkit
[11:39] Ubit Umarov: that is a version actually made for .net4.6
[11:40] Ubit Umarov: and not the more recent crap that in same cases try to add 100 dlls
[11:40] Ubit Umarov: to make .net4 to look like .net5
[11:40] Ubit Umarov: in fact in same cases useing very dangerus hacks
[11:40] Andrew Hellershanks: Ubit, No. I'm on 0.9
[11:40] Andrew Hellershanks: git master.
[11:41] Ubit Umarov: so i went to like 2018 version of 2019
[11:41] Ubit Umarov: still .netframework clean..
[11:41] Ubit Umarov: all changes them made after where basicly converting to .net5 as i said
[11:42] Ubit Umarov: ohhh it i 2020
[11:42] Ubit Umarov: it is..
[11:42] Andrew Hellershanks: Yes, the new MailKit based mail module was added on the day of last weeks meeting not long after Ubit had been telling me how bad it was and that I should not be using it. :P
[11:42] Ubit Umarov: 2.9.0    1,937,339    9/12/2020
[11:43] Ubit Umarov: yes i had look to recent versions
[11:43] Ubit Umarov: then as i just said went to the 2020 version
[11:43] Ubit Umarov: that is pure .net framework
[11:44] Ubit Umarov: well needs anohter one for ssl, but that is also clean .net framework
[11:44] Ubit Umarov: but this only adds tls
[11:45] Ubit Umarov: to actuall have a working llEmail people need to run own mailler
[11:45] Ubit Umarov: gmail will just stop working in May
[11:46] Ubit Umarov: others still accept simple tls connections, but do heavy checks on sender
[11:46] Andrew Hellershanks: When I was looking at MailKit it offered more than just TLS, IIRC.
[11:46] Ubit Umarov: so basicly only one machine can send
[11:47] Ubit Umarov: even so in same cases you need to go look to your mobile phone and then go a web site to put codes
[11:47] Ubit Umarov: email is getting totally outside the use of applications like opensim
[11:48] Ubit Umarov: so only grids, running own postifix or similar, with full credentials set can actually send emails
[11:49] Ubit Umarov: and we never had llemail for prims
[11:49] Ubit Umarov: only inside same region
[11:50] Ubit Umarov: so it only words sending messages to prims on same region, or with many restrictions, emails to humans
[11:50] George Equus is on
[11:50] Ubit Umarov: and very easy a script totally trigger maillers anti-spam thing,s and totally block a account
[11:52] Ubit Umarov: with this llEmail, should just be unsupported, except by grids having the extra needed support
[11:52] Ubit Umarov: but well the modules are there, and still looking..
[11:53] Ubit Umarov: err stll 2 modules the EmailModule and EmailModule2
[11:54] Ubit Umarov: i will merge them soon
[11:54] Ubit Umarov: EmailModule is the one we had so far without tls
[11:54] Ubit Umarov: and yes andrew
[11:55] Ubit Umarov: mailkit does a lot more than tls, and all useless
[11:55] Andrew Hellershanks: We are at almost 5 minutes to the top of the hour. Does anyone have a question or comment to make before the hour is up?
[11:55] Ubit Umarov: only feature actually usefull for most maillers is just that ssl/tls
[11:55] Michael.Christopher i guess not this week
[11:55] Andrew Hellershanks: Ubit, I don't think it was all useless. You can't predict what some external mail system will want from a sender.
[11:57] Ubit Umarov: for special ones, you need a special module anyways.. so..  whatever
[11:57] Andrew Hellershanks: The barmade has become invisible or we are haunted by the ghost of a barmade. :)
[11:57] Ubit Umarov: btw gmail and others will not work, because move to OAuth
[11:58] Ubit Umarov: and that is almost one set of code for each..
[11:58] Ubit Umarov: and requere a application registration.. with a clientID and clientSecret
[11:58] Ubit Umarov: not sure how that can work with a opensource app like opensim
[11:59] Ubit Umarov: also needs websites for user to confirm things..  etc
[11:59] Ubit Umarov: a total useless mess
[11:59] Ubit Umarov: i may actully kill my gmail account
[12:00] Ubit Umarov: i just have no time for the stupidity of confirmations it now wants
[12:00] Andrew Hellershanks: I have a family member who uses gmail. I don't know what they will do when that gmail changes their system. I don't use gmail so that is unlikely to affect me.
[12:00] Ubit Umarov: yes gavins usa changed time at strange moments...
[12:01] Ubit Umarov: for web access, it works..
[12:01] Gavin.Hird that did not register with me :-)
[12:01] Michael.Christopher hey Gavin
[12:01] Andrew Hellershanks: Hello, Gavin.
[12:01] Gavin.Hird Hi all
[12:01] Ubit Umarov: if you have the mobile phone at and to receive tokens
[12:01] Ubit Umarov: at hand..
[12:01] Andrew Hellershanks: Not everyone has a mobile phone.
[12:01] Ubit Umarov: fun
[12:02] Ubit Umarov: i can't do anything on my bank site without the mobile phone
[12:02] Gavin.Hird is voice down or is it just me?
[12:02] Ubit Umarov: stupid 2 factor auth means that
[12:02] Andrew Hellershanks: Gavin, I see white dots above heads so it may be just at your end.
[12:02] Ubit Umarov: i have voice always off
[12:03] Kayaker Magic: Welcome to the end of the meeting Gavin! California is an hour early this week.
[12:03] Ubit Umarov: no idea
[12:03] Ubit Umarov: well let me see more code changes
[12:03] Ubit Umarov: not much
[12:03] Gavin.Hird probably - compiled the viewer with an new version of Xcode and catastrophe ensues...
[12:03] Kayaker Magic: I heard Selby on voice
[12:03] Jagga Meredith: gotta go catch cat.  wish me luck.
[12:03] Andrew Hellershanks: ok, Jagga. Hope it goes well.
[12:03] Ubit Umarov: a few cosmetics changes, like the indentation etc
[12:04] Ubit Umarov: some "useless ones" like save a bit of cpu time  here and there
[12:04] Andrew Hellershanks: Ubit has been busy making code changes this week. He must be on March break. :)
[12:05] Michael.Christopher i've got to head out again, but have a great week everyone!@
[12:05] Ubit Umarov: changed how getdisplaynames work to save some useless copis of strings
[12:05] Selby.Evans Bye all
[12:05] Ubit Umarov: cya
[12:05] Andrew Hellershanks: Bye, Selby.
[12:06] Vincent.Sylvester Indented someones code and nuked all the tabs from orbit
[12:07] Ubit Umarov: removed strack trace on some Yengine script errors
[12:07] Ubit Umarov: well the out of heap ones
[12:07] Ubit Umarov: ahh on size of lists
[12:07] Ubit Umarov: made each strig char count as 2 bytes
[12:07] Ubit Umarov: it was counting as one so far
[12:08] Ubit Umarov: it also counts the pointers
[12:08] Ubit Umarov: that is 8 bytes per entry on the list
[12:08] Vincent.Sylvester Cleared or fixed a ton of mantis tickets too which will hopefully continue at a good pace to clear some of the backlog out, though the level of testing required for some of them means it'll be a while
[12:08] Ubit Umarov: ( 4 in 32 bit mode )
[12:09] Andrew Hellershanks: Ubit, where is that string char count used?
[12:09] Ubit Umarov: so your big scripts may get out of memory now :)
[12:10] Ubit Umarov: script memory usage, for example, andrew
[12:10] Ubit Umarov: is byte counts.. was not
[12:11] Andrew Hellershanks: Can't it detect the char size or are scripts always saved with two bytes per char?
[12:11] Ubit Umarov: .net char is utf-16 ie 2bytes
[12:11] Andrew Hellershanks: I thought there wasn't a limit on script size with OS.
[12:12] Ubit Umarov: Y does limit heap and stack use per script
[12:12] Andrew Hellershanks: I hope this won't break one of my scripts that is 3,000 lines long.
[12:12] Ubit Umarov: X did only limit stack, blowing up a violating script
[12:13] Ubit Umarov: there are ini entries to set the limits
[12:13] Ubit Umarov: well and this limits are old info
[12:13] Vincent.Sylvester This was adjusted after the whole heap size thing that was regarding a mantis ticket, which I still don't fully comprehend, but is evidently out of spec for what reasonable limits we do have
[12:13] Ubit Umarov: the onlly dif is fix on strings byte len, as i said
[12:14] Vincent.Sylvester If you are running out of heap you are doing something very wrong or are trying to implement a kernel in LSL
[12:14] Ubit Umarov: well it is easy to run out of defautl heap now
[12:14] Vincent.Sylvester Doing it on purpose doesn't count lol
[12:14] Ubit Umarov: nahh
[12:15] Andrew Hellershanks: :)
[12:15] Ubit Umarov: ofc Xengine did let your script send all memory to swap
[12:15] Ubit Umarov: and also fill swap :p
[12:16] Ubit Umarov: yeha there was a mantis, but that jsut made me look
[12:16] Ubit Umarov: the mantis was bad
[12:17] Vincent.Sylvester Half the stuff I been closing are things that just lack information making it near impossible to figure out what exactly is wrong
[12:17] Ubit Umarov: they wehre tring to detect max allowed heap by filling it on Xengine
[12:17] Andrew Hellershanks: That's nasty.
[12:17] Ubit Umarov: that could mean they where sending the entire box into swap
[12:18] Ubit Umarov: ofc Yengine does have getfreememory
[12:18] Ubit Umarov: and used
[12:19] Ubit Umarov: ll..
[12:20] Ubit Umarov: sad this llemail thing
[12:20] Ubit Umarov: i may try to fix it for internal prims use
[12:20] Ubit Umarov: it was a fail to do it region side
[12:20] Ubit Umarov: it must be grid side
[12:20] Ubit Umarov: a simple c# mailler
[12:21] Ubit Umarov: that is also the only poit that tries to send to a external mailler
[12:21] Ubit Umarov: to it can control quotas etc
[12:21] Ubit Umarov: and not let it jsut get blocked
[12:22] Ubit Umarov: and take email to prims totally out
[12:22] Ubit Umarov: just i do not like the http pooling  reading those may require
[12:22] Ubit Umarov: pooling just does not scale well with grid size
[12:24] Ubit Umarov: well a bit pass the hour no?
[12:24] Ubit Umarov: :)
[12:24] Andrew Hellershanks: Just a bit. :)
[12:25] Andrew Hellershanks: Time to wrap it up for this week.
[12:25] Andrew Hellershanks: Thank you all for coming. See you again next week.
Personal tools
About This Wiki