« Réunion du 01-02-2022 » : différence entre les versions

De OSWiki
Aller à la navigation Aller à la recherche
Aucun résumé des modifications
Ligne 2 : Ligne 2 :
=Introduction et cache des objets=
=Introduction et cache des objets=
* [[Chat_log_from_the_meeting_on_2022-01-25/fr#Viewers_:_Cache_des_objets_.28suite|Cache des objets]]
* [[Chat_log_from_the_meeting_on_2022-01-25/fr#Viewers_:_Cache_des_objets_.28suite|Cache des objets]]
 
* [https://www.firestormviewer.org/choose-your-platform/ Viever Firestorm]
* [https://vcs.firestormviewer.org/phoenix-firestorm/summary Dépôt des sources de Firestorm]
<pre>
<pre>
[11:02] Kayaker Magic murmure : Bonjour à tous !
[11:02] Kayaker Magic murmure : Bonjour à tous !
Ligne 23 : Ligne 24 :
[11:15] Ubit Umarov : je pense que cela réduit aussi le flot de commentaires.
[11:15] Ubit Umarov : je pense que cela réduit aussi le flot de commentaires.
</pre>
</pre>
=Miroir=
=Miroir=
<pre>
<pre>

Version du 2 février 2022 à 12:10

Source : http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2022-02-01

Introduction et cache des objets

[11:02] Kayaker Magic murmure : Bonjour à tous !
[11:03] Gavin.Hird @grid.xmir.org:8002 : Salut Andrew, Kayaker
[11:06] Vincent.Sylvester @hg.zetaworlds.com:8002 : La bêta de FS étant sortie, il semble que je puisse enfin fermer cette mantis sur le cache d'objets du viewer, c'est une bonne nouvelle.
[11:07] Andrew Hellershanks : Les notes de version font-elles mention d'un changement dans la gestion du cache objets ?
[11:07] Vincent.Sylvester @hg.zetaworlds.com:8002 : Etant donné que c'était le code principal de SL et qu'ils construisent habituellement par dessus, je suppose que c'est là maintenant.
[11:08] Vincent.Sylvester @hg.zetaworlds.com:8002 : Je suppose que leur tracker a toujours un ticket ouvert à ce sujet, mais cela pourrait changer quand la version complète sera disponible.
[11:08] Selby.Evans @grid.kitely.com:8002 : bonjour tout le monde
[11:09] Michael.Christopher @cc.opensimulator.org:8002 : bonjour à tous :)
[11:09] Gavin.Hird @grid.xmir.org:8002 : Salut Selby, Michael
[11:09] Jamie.Jordan @grid.kitely.com:8002 : Bonjour à tous
[11:09] Andrew Hellershanks : Bonjour, Selby, Michael et Jamie.
[11:10] Michael.Christopher @cc.opensimulator.org:8002 : bon de vous voir tous, longue semaine lol
[11:11] Vincent.Sylvester @hg.zetaworlds.com:8002 : Je n'ai même pas encore lu entièrement l'email, il y a toujours une tonne de texte dedans, je pense qu'il y avait d'autres corrections et de nouvelles fonctionnalités aussi, j'ai vu quelque chose à propos de la gestion de la "voice".
[11:13] Vincent.Sylvester @hg.zetaworlds.com:8002 : Accumulation de travail hein. Heureusement, il n'y a pas beaucoup de changements dans OpenSim cette semaine, juste des cosmétiques et la refactorisation de certains codes.
[11:14] Ubit Umarov : l'email n'est pas censé être une liste de changements importants.
[11:14] Ubit Umarov : et l'email signifie qu'il est en phase de bêta test (presque) fermé ?
[11:15] Vincent.Sylvester @hg.zetaworlds.com:8002: Je suppose qu'on attend de nous de trouver certains problèmes.
[11:15] Ubit Umarov : je pense que cela réduit aussi le flot de commentaires.

Miroir

[11:15] Ubit Umarov : oh joli miroir incurvé :)
[11:16] Andrew Hellershanks : Un miroir fonctionnel ?
[11:16] Kayaker Magic : Non, c'est la nouvelle fonction de photo à 360°, enroulée autour d'une sphère.
[11:16] Andrew Hellershanks: ah
[11:16] Michael.Christopher @cc.opensimulator.org:8002 : sympa.
[11:17] Ubit Umarov : bien pas un miroir... on voit vincent dessus et vincent est derrière.
[11:17] Andrew Hellershanks : Bel effet.
[11:17] Ubit Umarov : un snap seul ne peut pas faire un miroir.
[11:17] Kayaker Magic : Je pense que je ne l'ai pas inversé correctement, il y a un paramètre de texture pour le faire correctement.....
[11:17] Ubit Umarov: it needs the, well mirror, operatiion
[11:18] Ubit Umarov: mb flip on h
[11:18] Kayaker Magic: First thing I did to test the beta version of FS!
[11:18] Ubit Umarov: i only test basic.. be here :)
[11:18] Ubit Umarov: i just instaled it
[11:19] Ubit Umarov: so that 350 worked without region changes?
[11:19] Ubit Umarov: 360 also
[11:19] Kayaker Magic whispers: I'm running it on an Ubuntu desktop, always amazed there is a viewer that works on linux!
[11:19] Michael.Christopher @cc.opensimulator.org:8002: :)
[11:20] Ubit Umarov: oh i see the 360 there also
[11:20] Ubit Umarov: not that fan.. but ok. its there
[11:20] Kayaker Magic: In the Avatar menu
[11:20] Vincent.Sylvester @hg.zetaworlds.com:8002: I tried that in the past using camera rotation script to make a 180deg panorama, bit of a pain to stitch that together
[11:20] Ubit Umarov: there is also the tune thing
[11:21] Ubit Umarov: shows some preformance information for nerds and alike
[11:21] Ubit Umarov: also had a auto tune mode
[11:22] Ubit Umarov: experimental things
[11:22] Ubit Umarov: some more info at Beq's blog https://beqsother.blogspot.com/2021/12/upgraders-of-lost-arc.html
[11:24] Ubit Umarov: well on opensim code, only a few changes
[11:24] Ubit Umarov: that i do not remember.. hmm hmm lets see
[11:24] Ubit Umarov: oh on take copy, make the server tell when it is refused
[11:25] Ubit Umarov: for normal grid users the veriwers are just told to not show the menu if the user as no rights
[11:25] Ubit Umarov: but on HG, if export to HG is blocked, it is shown.. so, confusing
[11:26] Ubit Umarov: the way hg is done is hard to add that check inlike so tell viewers to not show it
[11:27] Ubit Umarov: on that case, removed the block of export of landmarks
[11:27] Ubit Umarov: not nice to do that...
[11:28] Vincent.Sylvester @hg.zetaworlds.com:8002: It seems some want it all, HG everywhere, but only on strict terms defeating the entire open nature of it, which is a bit silly
[11:28] Ubit Umarov: made also some changes on HG instant message modules
[11:29] Ubit Umarov: tried to reduce the spargetti convolution
[11:29] Vincent.Sylvester @hg.zetaworlds.com:8002: Broke a bunch of my patches I put in there to debug some of the HG IM failures that happen from time to time
[11:30] Ubit Umarov: there was a scare reentrant call,  i removed it.. the crashes are on same stack level
[11:30] Ubit Umarov: and not Stack Overflow, that older code could do
[11:31] Ubit Umarov: in fact got one at LBSA on a intermediate code change during last week :)
[11:32] Ubit Umarov: well HG IMs and IMs in general, still very heavy and confuding code :(
[11:32] Ubit Umarov: sad, bc IMs are a crucial component, ofc
[11:32] Michael.Christopher @cc.opensimulator.org:8002: ofc
[11:33] Ubit Umarov: no idea why the well tested IRC model was not used
[11:33] Ubit Umarov: instead of al this twisted thing
[11:33] Vincent.Sylvester @hg.zetaworlds.com:8002: I been adding debug code to that stuff since I get reports of it not working right and I think there might be something in the current configs that misleads people during configuration causing that stuff to always fail, but I am still not sure what exactly that is
[11:33] Ubit Umarov: ofc HG makes it even more confusing and slower
[11:34] Michael.Christopher @cc.opensimulator.org:8002: i was sitting here pondering the feasibility of an independent chat module to make cross-platform chat easier; hg IMs should be easier that out-of-world i would think
[11:34] Ubit Umarov: and it is just impossible to make HG IMs work in all cases
[11:35] Andrew Hellershanks: I was wondering how much work it would be to change the IM system to be more like the IRC model.
[11:35] Ubit Umarov: friends is the only case it can kinda work
[11:37] Ubit Umarov: i think some grids even used modified IRCDs back in the day
[11:37] Ubit Umarov: but never seen one
[11:37] Gavin.Hird @grid.xmir.org:8002: since we are talking about performance, there was an interesting revelation at the latest TPV meeting
[11:37] Ubit Umarov: yes gavin?
[11:38] Andrew Hellershanks: Sometimes it feels like a lot of the code should be rewritten to clean it up and have it work properly.
[11:38] Michael.Christopher @cc.opensimulator.org:8002: but that makes sense to me - you'd want to meet someone before giving them your phone, so only allowing hg ims from friends
[11:38] Gavin.Hird @grid.xmir.org:8002: It turns out the FS area search fuction does a select all in the background on all objects in the region
[11:38] Gavin.Hird @grid.xmir.org:8002: and then search the hover info for display in the floater
[11:39] Gavin.Hird @grid.xmir.org:8002: so if you have  a region with many objects you get a request storm for the simulator
[11:39] Michael.Christopher @cc.opensimulator.org:8002: ....is that really the cleanest way they could get it to work?
[11:39] Ubit Umarov: uff
[11:39] Gavin.Hird @grid.xmir.org:8002: aparently there is no other way to get that info as it is
[11:39] Andrew Hellershanks: That seems odd, Gavin.
[11:39] Kayaker Magic: You expect them to do things in a clean or logical way???
[11:40] Andrew Hellershanks: Couldn't they get the data needed from the scene data?
[11:40] Michael.Christopher @cc.opensimulator.org:8002: lol fair enough
[11:40] Ubit Umarov: fs not and then as strange pre caching issues
[11:40] Gavin.Hird @grid.xmir.org:8002: LL and the FS people did not think so
[11:40] Ubit Umarov: now and then
[11:40] Gavin.Hird @grid.xmir.org:8002: there was a discussion on adding server side functionality
[11:41] Ubit Umarov: a a point moving mouse over horizon with diferent parcels made it ask for parcels information on almost every mluse tick
[11:41] Ubit Umarov: a huge flood
[11:41] Michael.Christopher @cc.opensimulator.org:8002: ouch
[11:41] Andrew Hellershanks: not good
[11:41] Ubit Umarov: ohh fs also likes to ask serveral things 3 times.. always..
[11:42] Ubit Umarov: 1 to 5ms time dif, same udp serial number..  yeack
[11:42] Jagga Meredith: what I tell you 3 times is true
[11:43] Michael.Christopher @cc.opensimulator.org:8002: so its a performance vs security question on the x3, not a glitch
[11:43] Vincent.Sylvester @hg.zetaworlds.com:8002: Having db access for that is just a lot more useful especially if you need to find stuff across multiple regions
[11:43] Ubit Umarov: no it is a plain bug somewhere
[11:43] Ubit Umarov: just that
[11:43] Ubit Umarov: :p
[11:44] Michael.Christopher @cc.opensimulator.org:8002: oh ok lol
[11:44] Vincent.Sylvester @hg.zetaworlds.com:8002: Suppose could add a cap to get the data back as a nicely formatted xml structure xD
[11:44] Ubit Umarov: or was... some time since i check
[11:44] Andrew Hellershanks: Makes me think of the Space Shuttle computers. There are three of them. If one disagrees with the other two it is ignored.
[11:45] Andrew Hellershanks: Vincent, Don't say XML too loudly. ;)
[11:45] Ubit Umarov: so that amstrong had to kill all them and land by hand :P
[11:45] Michael.Christopher @cc.opensimulator.org:8002: rofl
[11:46] Jagga Meredith: I was quoting Hunting of the Snark
[11:46] Andrew Hellershanks: Jagga, I don't know that. Is it a book?
[11:46] Vincent.Sylvester @hg.zetaworlds.com:8002: Doing a select of prims for 58k prims takes .3 seconds according to exec runtime of my bash script, guess the query is cached huh
[11:47] Jagga Meredith: poem by Lewis Carrollof Alice in Wonderland fame
[11:47] Ubit Umarov: oh some houses around total stop fs on select
[11:47] Andrew Hellershanks: Jagga, ah, ok.
[11:47] Ubit Umarov: like 1k linked prims
[11:48] Michael.Christopher @cc.opensimulator.org:8002: does sound a lot like the DDNS I think we need too - if each OS server talks to a couple others regularly...
[11:48] Ubit Umarov: adn select is also a big storm on regions
[11:48] Ubit Umarov: and not that light code on them
[11:48] Ubit Umarov: they need to disable/enable some physics etc etc on select/de
[11:48] Ubit Umarov: very bad protocol thing still
[11:50] Ubit Umarov: so what news do you have abotu opensim?
[11:50] Ubit Umarov: ..what more..
[11:51] Andrew Hellershanks: Nothing more in the way of news other than what has already been mentioned.
[11:52] Ubit Umarov: ohh and we are in Feb and i did not release 0.9.2.1 i was planning for Jan
[11:52] Andrew Hellershanks: Just under 10 minutes remaining before the top of the hour. Any questions or comments?
[11:52] Michael.Christopher @cc.opensimulator.org:8002: not here
[11:52] Ubit Umarov: also not that many changes to jsutify it
[11:53] Andrew Hellershanks: Ubit probably didn't flip over the page on his calendar so he things it is still January. ;)
[11:53] Ubit Umarov: i just noticed during login here :p
[11:53] Gavin.Hird @grid.xmir.org:8002: I mentioned last time that the latest update of Windows clobbered my ability to test the Windows viewer in vmware
[11:54] Gavin.Hird @grid.xmir.org:8002: I managed to work around that, but it is worse
[11:54] Ubit Umarov: i can release at  Jan(31+12)th 2022
[11:54] Gavin.Hird @grid.xmir.org:8002: much worse
[11:55] Ubit Umarov: :(
[11:55] Jagga Meredith: I had probtlem with red could not connect to database messages that survived multiple reboots, but magically fixed itself after some manual db queries.  I'm running mariadb.  Anybody else notice anytrhing?
[11:55] Gavin.Hird @grid.xmir.org:8002: when rebuilding the latest version with no changes in the updated Window, it produce a viewer that randomly manages to load mes or textures
[11:55] Gavin.Hird @grid.xmir.org:8002: completely unpredictable
[11:55] Ubit Umarov: is vmware also losing momentum, turning into obsolete ?
[11:56] Gavin.Hird @grid.xmir.org:8002: so there must be some system lib in windows that it links with that now breaks the existing libs
[11:56] Andrew Hellershanks: Ubit, release it soon. It may still be January somewhere near the international date line. :)
[11:56] Ubit Umarov: ms seems very good on causing that all around it
[11:56] Jagga Meredith: thats what I was wonderinhg
[11:56] Gavin.Hird @grid.xmir.org:8002: this happens regardless if the viewer is built on a real windows system or a virtual systme
[11:57] Ubit Umarov: strange
[11:57] Gavin.Hird @grid.xmir.org:8002: so in short, the Windows viewer is fucked (excuse my french) as far as building it
[11:57] Ubit Umarov: and no idea on what kind of change?
[11:57] Gavin.Hird @grid.xmir.org:8002: I have no idea
[11:57] Ubit Umarov: ie what breaks?
[11:57] Jagga Meredith: mine started after a windows update
[11:58] Gavin.Hird @grid.xmir.org:8002: it randomly manages to load meshes or textures
[11:58] Gavin.Hird @grid.xmir.org:8002: completely different result every login
[11:58] Andrew Hellershanks: Don't know about vmware. I tried to use it a long time ago. I've been using virtualbox. Found it easier to get started with it.
[11:58] Ubit Umarov: outch
[11:58] Gavin.Hird @grid.xmir.org:8002: regadless of clean cache or not
[11:58] Ubit Umarov: that is binary llsd
[11:59] Andrew Hellershanks: Jagga, that is an odd problem. I haven't been using mariadb on my machine and the one where I do use it hasn't had that issue.
[11:59] Ubit Umarov: don't thing they can change endness on intel :)
[11:59] Vincent.Sylvester @hg.zetaworlds.com:8002: What is it loading them with, udp or http?
[11:59] Gavin.Hird @grid.xmir.org:8002: http
[12:00] Vincent.Sylvester @hg.zetaworlds.com:8002: If that's not available shouldn't it fall back on udp? Does that work more reliably?
[12:00] Gavin.Hird @grid.xmir.org:8002: no
[12:00] Andrew Hellershanks: Jagga, that's the problem. Never update software. It just seems to break things. That's the feeling in my household as it is often true.
[12:01] Gavin.Hird @grid.xmir.org:8002: it is not really feasible to revert the Windows environment either, because then the insane update policy of MS force you to either be completely disconnected from the interne, making it impossible to test properly, or it will force update you again
[12:02] Ubit Umarov: and acording to conspiracy teories ms just breaks some things intescionaly to promote new prodtcuts
[12:02] Ubit Umarov: intencionaly..
[12:02] Gavin.Hird @grid.xmir.org:8002: there was the old "DOS ain't done till Lotus won't run" in the 80s
[12:03] Ubit Umarov: netscape..
[12:03] Jagga Meredith: damn I miss Lotus
[12:03] Jamie.Jordan @grid.kitely.com:8002: I'm not upgrading til they force me
[12:03] Ubit Umarov: etc
[12:03] Gavin.Hird @grid.xmir.org:8002: Lotus 123 was a spreadsheet competitor to some MS product
[12:03] Vincent.Sylvester @hg.zetaworlds.com:8002: So something in http handling must have changed for it to randomly decide not to fire the request or something like that
[12:03] Jagga Meredith: appole's Visicalc
[12:04] Gavin.Hird @grid.xmir.org:8002: it could be the boos coroutine library that errs alos
[12:04] Gavin.Hird @grid.xmir.org:8002: also
[12:04] Andrew Hellershanks: I used to work for someone that knew the person who created Visicalc.
[12:04] Gavin.Hird @grid.xmir.org:8002: boost
[12:04] Ubit Umarov: gmm gzip gavin?
[12:04] Jagga Meredith: damn we're old
[12:04] Gavin.Hird @grid.xmir.org:8002: as the textures are fetched with a couroutine
[12:04] Andrew Hellershanks: hehe
[12:05] Gavin.Hird @grid.xmir.org:8002: gzip?
[12:05] Ubit Umarov: mesh had a issue with header bytes on unzip
[12:05] Ubit Umarov: libs used by opensim had to hack ignore like 4 bytes
[12:05] Ubit Umarov: on mesh unzip
[12:07] Ubit Umarov: using (DeflateStream decompressionStream = new DeflateStream(inMs, CompressionMode.Decompress))
                            {
                                byte[] readBuffer = new byte[8192];
                                inMs.Read(readBuffer, 0, 2); // skip first 2 bytes in header
                                int readLen = 0;

                                while ((readLen = decompressionStream.Read(readBuffer, 0, readBuffer.Length)) > 0)
                                    outMs.Write(readBuffer, 0, readLen);
[12:07] Ubit Umarov: still do
[12:07] Ubit Umarov: bc viewers encoder
[12:07] Jagga Meredith: ewww
[12:08] Gavin.Hird @grid.xmir.org:8002: if there was a zlib issue, nothing would arrive really
[12:08] Ubit Umarov: ( that now uses .net native gzip )
[12:09] Ubit Umarov: yeah.,. well this is a old issue i remember, on meshes
[12:09] Ubit Umarov: hack above by dahlia or lkalif..
[12:09] Ubit Umarov: ( i think )
[12:09] Gavin.Hird @grid.xmir.org:8002: so I am working on a plan B
[12:09] Andrew Hellershanks: Skip first two bytes? Is that to get past the magic string? That is usually four bytes.
[12:10] Ubit Umarov: guess that is explained on some libopenmetaverse or opensim emails :)
[12:10] Ubit Umarov: or commits
[12:11] Ubit Umarov: i remember i did knew why.. but forgot the why :P
[12:12] Andrew Hellershanks: Ubit, anything helpful when using git blame to track down when that line was last touched that would help track down the reason?
[12:12] Ubit Umarov: possible
[12:12] Ubit Umarov: but gavin just told its something else
[12:12] Ubit Umarov: fun debug for him :)
[12:13] Gavin.Hird @grid.xmir.org:8002: that is easy for you to day
[12:13] Gavin.Hird @grid.xmir.org:8002: to say*
[12:13] Ubit Umarov: ofc :p
[12:13] Michael.Christopher @cc.opensimulator.org:8002: wish I could stasy, but I've got to go pick my son up from school.  Have a great week, talk to you soon, email me if you need me!
[12:14] Ubit Umarov: and fun means "fun" ofc :)
[12:14] Ubit Umarov: cya Michael
[12:15] Ubit Umarov: wel good reasons why software updates do scare me
[12:15] Selby.Evans @grid.kitely.com:8002: must go --bye all
[12:15] Ubit Umarov: one can't ever know what will break
[12:15] Ubit Umarov: cya Selby.Evans
[12:15] Ubit Umarov: ( one reason why i hate nuget and its keep updated crap )
[12:16] Andrew Hellershanks: Its is now quarter past the hour. If there are no last minute items for today I'll wrap up the meeting.
[12:17] Jamie.Jordan @grid.kitely.com:8002 whispers: I need to step out have a great week guys
[12:17] Ubit Umarov: cya jamie.Jordan
[12:17] Andrew Hellershanks: ok, Jamie. Thanks for stopping by.
[12:17] Andrew Hellershanks: All righty then. That will do it for another week. Thank you all for coming. See you again next week.
[12:17] Ubit Umarov: so quiet there motoko :)
[12:17] Jagga Meredith: godd meeting.  Got so deep in weeds I'm covered in burrs.
[12:17] Kayaker Magic: Thanks and bye all!