Réunion du 26-07-2022

Source : http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2022-07-26

Introduction

[11:01] Jamie.Jordan @grid.kitely.com:8002 : Bonjour à tous

[11:02] Andrew Hellershanks : Bonjour, tout le monde.

[11:02] Kayaker Magic : Je ne suis pas très content d'Ubuntu : Les pilotes de son sont terribles, le plug-and-pray n'arrête pas de changer mes choix de périphériques de son, la configuration des icônes sur l'écran est une plaie.

[11:02] Andrew Hellershanks : Les pilotes de son posent problème depuis le début de Pulse Audio.

[11:02] Ubit Umarov : aucune idée, je n'ai pas de machine pour essayer un linux.

[11:03] Ubit Umarov : il y a environ 2 ans, j'avais debian 10 sur une machine, et ça marchait bien.

[11:03] Ubit Umarov : ensuite la machine est morte.

[11:03] Ubit Umarov : bahh

[11:03] Kayaker Magic : Oh et sur Ubuntu il y a un truc d'interface appelé Snap que la moitié des programmes utilisent, il crée un nouveau périphérique logique pour chaque application, remplissant ma liste de périphériques réels.

[11:04] Andrew Hellershanks : Kayaker, j'ai laissé tomber Ubuntu quand ils ont ajouté Unity pour leur bureau.

[11:05] Andrew Hellershanks: Unity? Hm... ce n'est peut-être pas le bon nom.

Changements dans OpenSim cette semaine

[11:05] Vincent.Sylvester @hg.zetaworlds.com:8002 : Pas beaucoup de changements dans OpenSim ces derniers temps. Des trucs mineurs dans libomv car nous sommes à la recherche de ce qui pourrait clocher dans le crash de TryParse. Bien sûr, en testant sur la dernière version de mono, aucune des régions qui plantaient auparavant ne l'a fait, donc je suis de retour à la case départ sur ce point. Je vais probablement revenir à la dernière version et voir ce qui se passe, il doit y avoir autre chose que des scripts qui causent cela, peut-être des gens.

[11:05] Kayaker Magic : J'ai une question à propos d'OpenSim : J'ai reçu une demande pour faire tourner OpenSim 0.8.2 sur une région. Je l'ai récupéré dans les archives, il plante et ne fonctionne pas. Probablement un problème avec mono. Quelqu'un a-t-il une idée de ce que je peux essayer ?

[11:05] Vincent.Sylvester @hg.zetaworlds.com:8002 : Sur quoi se bloque-t-il ?

[11:06] Kayaker Magic : Attends, je regarde ça.

[11:07] Andrew Hellershanks : Si tu utilises la 0.8.2 sur une région, quelle version d'OS les autres régions utilisent-elles ?

[11:08] Kayaker Magic : J'utilise un Yeti maître assez récent sur le reste, tous connectés à OSGrid.

[11:08] Kayaker Magic : Voilà l'erreur :

[11:08] Kayaker Magic:

2022-07-23 06:18:59,121 ERROR (Heartbeat-(Esses_Workshop)) - OpenSim.Region.Framework.Scenes.Scene [SCENE]: Failed on region Esses Workshop with exception Object reference not set to an instance of an object  at OpenSim.Region.Framework.Scenes.SceneGraph.UpdatePreparePhysics () [0x00006] in <35ba7167b3194ad9abe7c1bc9afcecb6>:0
 at OpenSim.Region.Framework.Scenes.Scene.Update (System.Int32 frames) [0x00140] in <35ba7167b3194ad9abe7c1bc9afcecb6>:0
2022-07-23 06:18:59,124 ERROR (Heartbeat-(Esses_Workshop)) - OpenSim.Application [APPLICATION]: ^M
APPLICATION EXCEPTION DETECTED: System.UnhandledExceptionEventArgs^M
^M

[11:09] Kayaker Magic : Utilisation de XEngine, ubODE

[11:09] Andrew Hellershanks : Je ne pense pas que ce message d'erreur soit très utile.

[11:09] Andrew Hellershanks : Kayaker, quelle version de mono utilises-tu ?

[11:10] Kayaker Magic : Mais c'est ce que je pense de tous les messages d'erreur d'opensim ! Aucun d'entre eux n'est très utile !

[11:10] Kayaker Magic : J'utilise mono 6.12.

[11:10] Andrew Hellershanks : Eh bien, certains d'entre eux sont plus utiles que d'autres.

[11:11] Andrew Hellershanks : Jusqu'où va-t-il dans le processus de démarrage de la région avant que l'erreur n'apparaisse ?

[11:11] Kayaker Magic : On dirait que ça va commencer, il est écrit "Started tracking thread Heartbeat".

[11:12] Kayaker Magic : La région est toute nouvelle, pas de prims, pas de scripts, une île ronde.

[11:12] Ubit Umarov : c'est quoi la version 0.8.2 ?

[11:12] Ubit Umarov : ;)

[11:12] Vincent.Sylvester @hg.zetaworlds.com:8002 : C'est tout ce qu'il y a ? Peux-tu coller les parties restantes et peut-être quelques lignes avant cela. As-tu essayé une autre combinaison de scripts et de physique ?

[11:13] Vincent.Sylvester @hg.zetaworlds.com:8002 : Les versions sont marquées dans git donc tu peux aussi compiler à partir des sources sous un mono plus récent pour voir si cela aide.

[11:13] Kayaker Magic : Voici quelques lignes (tout semble bon) avant le crash :

[11:13] Kayaker Magic:

2022-07-23 06:18:58,795 DEBUG (1) - OpenSim.Data.MySQL.MySQLSimulationData [REGION DB]: Loaded 0 objects using 0 prims
2022-07-23 06:18:58,801 DEBUG (1) - OpenSim.Data.MySQL.MySQLSimulationData [REGION DB]: Loaded inventory from 0 objects
2022-07-23 06:18:58,802 INFO  (1) - OpenSim.Region.Framework.Scenes.Scene [SCENE]: Loaded 0 objects from the datastore
2022-07-23 06:18:58,803 DEBUG (1) - OpenSim.Region.CoreModules.Framework.UserManagement.UserManagementModule [USER MANAGEMENT MODULE]: Caching creators' data from Esses Workshop (0 objects)...
2022-07-23 06:18:58,806 INFO  (1) - OpenSim.Region.Framework.Scenes.Scene [SCENE]: Loading land objects from storage
2022-07-23 06:18:58,932 DEBUG (1) - OpenSim.Services.GridService.GridService [GRID SERVICE]: Region Esses Workshop (9b145e2a-5c81-4803-ab84-381904b61f69, 256x256) registered at 2294,9495 with flags RegionOnline
2022-07-23 06:18:59,114 INFO  (1) - OpenSim.Region.Framework.Scenes.Scene [SCENE]: Initializing script instances in Esses Workshop

[11:13] Kayaker Magic:

2022-07-23 06:18:59,114 INFO  (1) - OpenSim.Region.Framework.Scenes.Scene [SCENE]: Initialized 0 script instances in Esses Workshop
2022-07-23 06:18:59,115 DEBUG (1) - OpenSim.Framework.Monitoring.Watchdog [WATCHDOG]: Started tracking thread Heartbeat-(Esses_Workshop), ID 42

[11:13] Andrew Hellershanks : Si tu m'avais demandé la semaine dernière, j'aurais encore une copie de la 0.8.2 sur ma machine pour l'essayer car j'ai mono 6.12 installé.

[11:14] Kayaker Magic : Non, je n'ai pas essayé YEngine et autres combinaisons....

[11:14] Kayaker Magic : Y avait-il un YEngine à l'époque ?

[11:15] Andrew Hellershanks : Pas de YEngine dans la 0.8.2

[11:15] Ubit Umarov : non

[11:15] Vincent.Sylvester @hg.zetaworlds.com:8002: Well it crashes on physics so trying Bullet would be a start [11:16] Vincent.Sylvester @hg.zetaworlds.com:8002: Beyond that I think that version was still compiled on mono 2.10 so that's a big jump [11:16] Vincent.Sylvester @hg.zetaworlds.com:8002: When in doubt compile from source [11:16] Kayaker Magic: I tried compiling from source, got 21 warnings, then ran that copy, same error [11:17] Vincent.Sylvester @hg.zetaworlds.com:8002: Heartbeat fail should not cause application termination unless there is something else wrong [11:18] Andrew Hellershanks: You have double checked the configuration of the region to make sure it isn't a typo in an ini file? [11:18] Kayaker Magic: If I switch to Yeti master, it comes up fine with an almost identical INI file. [11:20] Vincent.Sylvester @hg.zetaworlds.com:8002: Compile in debug so it gives you line numbers and then start directly so you can fetch the full Stacktrace to see where it goes wrong, the log errors are not always descriptive enough [11:20] Andrew Hellershanks: It is the "Object reference not set to an instance of an object" error that is of concern. [11:21] Andrew Hellershanks: One thing that came to mind is a possible error in a database configuration string. [11:21] Kayaker Magic: There is a stack trace in the log after that error, should I dump in here? [11:22] Vincent.Sylvester @hg.zetaworlds.com:8002: The part under managed stacktrace that points to the code directly, skip the memory dump [11:23] Kayaker Magic: I mean more stuff in the log after that error message, here I will paste it: [11:23] Kayaker Magic: 2022-07-23 06:18:59,124 ERROR (Heartbeat-(Esses_Workshop)) - OpenSim.Application [APPLICATION]: ^M APPLICATION EXCEPTION DETECTED: System.UnhandledExceptionEventArgs^M ^M Exception: System.NullReferenceException: Object reference not set to an instance of an object

 at OpenSim.Region.Framework.Scenes.Scene.get_TimeDilation () [0x0000b] in <35ba7167b3194ad9abe7c1bc9afcecb6>:0
 at OpenSim.Region.Framework.Scenes.Scene.Update (System.Int32 frames) [0x0050e] in <35ba7167b3194ad9abe7c1bc9afcecb6>:0
 at OpenSim.Region.Framework.Scenes.Scene.Heartbeat () [0x0000c] in <35ba7167b3194ad9abe7c1bc9afcecb6>:0
 at System.Threading.ThreadHelper.ThreadStart_Context (System.Object state) [0x00014] in <de882a77e7c14f8ba5d298093dde82b2>:0
 at System.Threading.ExecutionContext.RunInternal (System.Threading.ExecutionContext executionContext, System.Threading.ContextCallback callback, System.Object state, System.Boolean preserveSyncCtx) [0x00071] in <de882a77e7c14f8ba5d298093dde82b2>:0
 at

[11:23] Kayaker Magic: System.Threading.ExecutionContext.Run (System.Threading.ExecutionContext executionContext, System.Threading.ContextCallback callback, System.Object state, System.Boolean preserveSyncCtx) [0x00000] in <de882a77e7c14f8ba5d298093dde82b2>:0

 at System.Threading.ExecutionContext.Run (System.Threading.ExecutionContext executionContext, System.Threading.ContextCallback callback, System.Object state) [0x0002b] in <de882a77e7c14f8ba5d298093dde82b2>:0
 at System.Threading.ThreadHelper.ThreadStart () [0x00008] in <de882a77e7c14f8ba5d298093dde82b2>:0 ^M

^M Application is terminating: True^M [11:24] Andrew Hellershanks: Are you building the code as a release? You need to run it compiled for debug mode to get a stack trace with tile and line numbers. [11:24] Andrew Hellershanks: Why does someone want to still us 0.8.2 if the rest of the regions are 0.9.x? Does the person have a script that won't run in 0.9? [11:25] Kayaker Magic: I thought the default build was debug. I just did ./runprebuild.sh and xbuild [11:26] Kayaker Magic: Person thinks that mesh uploads work different in 0.8, has not been receptive to "let me teach you the correct way to upload mesh" [11:26] Andrew Hellershanks: Ah, ok. [11:27] Vincent.Sylvester @hg.zetaworlds.com:8002: I would not even entertain that level of stupidity, enabling that is not the way to go [11:27] Kayaker Magic: Now that I think about it, she probably remembers how Bullet builds physics models for you. [11:28] Vincent.Sylvester @hg.zetaworlds.com:8002: So then upload bad mesh and pollute everything with potentially corrupt data causing other regions to crash even, fun [11:30] Andrew Hellershanks: There have been database migrations between 0.8.2 and 0.9 which might also be an issue. [11:30] Kayaker Magic: OK, that gives me a few ideas to try: Give her 0.9 with Bullet, force her to listen to mesh upload suggestions... [11:31] Kayaker Magic: Try bringing up her 0.8 with Bullet... Enough about this today. I have other questions! [11:31] Ubit Umarov: (at phone..) [11:31] Andrew Hellershanks: That would be a better idea thatn trying to get a 0.8 region running in a grid running mostly 0.9 [11:33] Andrew Hellershanks: Any other questions, Kayaker? [11:35] Kayaker Magic: I'm trying to figure out when things are written to the region assetcache. If I rez an object in a region, I don't see it in that cache. When does that happen? Only after a successful fetch from the main fsassets? [11:36] Andrew Hellershanks: That sounds like a qusetion for Ubit, once he is off the phone. [11:37] Kayaker Magic: And if you rez two copies of something, each one gets its own UUID, right? [11:39] Kayaker Magic: but each object has to have asset UUIDs for all its textures, are those cached in the assetcache by their asset UUIDs? [11:39] Andrew Hellershanks: yes to both. [11:39] Ubit Umarov: uuid is the identifier of any asset [11:40] Kayaker Magic: by asset UUID I mean the UUID in fsassets, is a texture cached in assetcashe by the same UUID? [11:42] Ubit Umarov: as i said the identifier of a asset is only its uuid [11:43] Kayaker Magic: I guess what I am asking is: If I look in the region database I see an object stored by its rezzed UUID, but is the only thing stored in the assetcache the atomic parts of it stored by the fsasset UUIDs? [11:44] Ubit Umarov: a object does lost uuids of the textures ( materials etc) it uses [11:44] Ubit Umarov: does list [11:45] Ubit Umarov: a object can be stored as a asset also, ofc [11:45] Ubit Umarov: that stacktrace is strange [11:45] Kayaker Magic: If I create a new object, say out of two cubes, is that object in the fsasset yet? WHen does it get saved the first time? [11:46] Ubit Umarov: something very worng with that region [11:46] Ubit Umarov: no a diference in objects is that are on the region db only ( own format ) [11:46] Vincent.Sylvester @hg.zetaworlds.com:8002: Prims are not assets, textures, mesh, animations, scripts, etc are assets. You cannot create an asset just rezzing a prim [11:47] Ubit Umarov: they turn into a asset where taken to inventory [11:47] Ubit Umarov: when.. [11:47] Vincent.Sylvester @hg.zetaworlds.com:8002: Yup [11:48] Ubit Umarov: just because implementation, in a region there are scene parts and scene object groups :) [11:48] Kayaker Magic: So if I rez a cube, nothing happens in fsassets or assetcache, if I put a new texture on it, the texture comes from fsassets, and is saved in assetcache. [11:48] Ubit Umarov: that have own storage on dbs [11:48] Ubit Umarov: yes [11:48] Kayaker Magic: OK, I think I'm getting the idea. [11:49] Ubit Umarov: on the oposite, textures are directly uploaded into grid ( fsassets) [11:49] Vincent.Sylvester @hg.zetaworlds.com:8002: That is if they are not textures from the library which each region supplies, so you'd probably have to use a texture that you uploaded yourself [11:49] Ubit Umarov: ( except dynamic ones ) [11:49] Kayaker Magic: I see [11:49] Ubit Umarov: so when we apply one to a face, viewer will ask for it [11:49] Ubit Umarov: vuia assets service [11:50] Ubit Umarov: via [11:50] Ubit Umarov: on 0.9.2 via ViewerAsset cap [11:50] Ubit Umarov: well since some 0.9 when i added that cap support .) [11:51] Ubit Umarov: objects also have own paths to viewer [11:51] Kayaker Magic: NP, I always use 0.9.x unless dragged kicking and screaming back... [11:51] Vincent.Sylvester @hg.zetaworlds.com:8002: object viewer cache, that fun thing [11:51] Ubit Umarov: special protocol to send "objects" to viewer as such.. [11:52] Ubit Umarov: well porotocols,.. object viewer cache uses other udp protocol, one could say [11:52] Vincent.Sylvester @hg.zetaworlds.com:8002: Yeah don't use old viewers you gonna have a bad time lol [11:53] Ubit Umarov: objects inside other objects are assets ofc [11:53] Ubit Umarov: btw fs did a release [11:53] Vincent.Sylvester @hg.zetaworlds.com:8002: primitems pointing at the assets [11:54] Kayaker Magic: FINALLY! [11:54] Ubit Umarov: not that many diferences [11:54] Vincent.Sylvester @hg.zetaworlds.com:8002: grid status button now works [11:54] Kayaker Magic: Well that Vivox fix was needed to prevent crashing when HG TPing between some grids. [11:54] Ubit Umarov: now library does use own caps [11:55] Ubit Umarov: i had added those caps like in 2018 then removed since no viewer used them [11:55] Ubit Umarov: alchemy asked me to put them back, so the worked on it, now also fs [11:55] Ubit Umarov: no big dif [11:56] Ubit Umarov: just when all viewers do it, region code can be simplified [11:56] Vincent.Sylvester @hg.zetaworlds.com:8002: Well slightly cleaner code too so that's good, so many areas really needing a review [11:56] Ubit Umarov: removing the hack of giving library on the normal inventory caps [11:57] Ubit Umarov: no idea if i missed any question [11:57] Ubit Umarov: phone had to ring jsut at start of meeting, ofc [11:58] Kayaker Magic: I got usefull replies to my two questions. [11:58] Kayaker Magic: Hogging a lot of time again! Thanks guys! [11:58] Ubit Umarov: ahh [11:58] Vincent.Sylvester @hg.zetaworlds.com:8002: You can hop on irc if you need help, don't have for the meeting to roll around [11:59] Ubit Umarov: i made several (useless) changes on ubode and did commit to git [11:59] Ubit Umarov: but currently that is kinda broken :p [11:59] Andrew Hellershanks: Vincent, but what would we talk about if all the questions are handled via IRC during the week? :) [11:59] Ubit Umarov: somehow the old gcc i was using generates code that fails to do raycasts [12:00] Ubit Umarov: compiled on win and debian11 gcc, works fine [12:00] Ubit Umarov: s oa pretty nice bug.. impossible to find [12:00] Ubit Umarov: gcc jsut went nuts [12:00] Ubit Umarov: duhhh [12:00] Vincent.Sylvester @hg.zetaworlds.com:8002: Where is the source for that? Been meaning to ask [12:01] Ubit Umarov: opensim-libs [12:01] Vincent.Sylvester @hg.zetaworlds.com:8002: Maybe I'll get a completely different cryptic error... knowing my luck [12:01] Andrew Hellershanks: Wow. What and odd sort of bug. [12:01] Ubit Umarov: wlel no error [12:01] Ubit Umarov: just raycasts fail [12:01] Ubit Umarov: no error ... nothing.. [12:02] Ubit Umarov: raycasts used internally to terrain collisions, kinda work till they don't :9 [12:02] Ubit Umarov: then the prim gets odd collisions [12:02] Vincent.Sylvester @hg.zetaworlds.com:8002: Didn't we have a problem with that a while back as well? [12:02] Ubit Umarov: gcc just lost some code path [12:03] Vincent.Sylvester @hg.zetaworlds.com:8002: It killing cpu [12:03] Ubit Umarov: a failed init, it used to do.. a lost reference.. no idea [12:03] Vincent.Sylvester @hg.zetaworlds.com:8002: Maybe it's optimizing it away [12:03] Ubit Umarov: on a recent gcc, same code.. no issues [12:04] Jamie.Jordan @grid.kitely.com:8002: Have a great week yall [12:04] Ubit Umarov: but i can't compile it on very last gcc, or many will no be able to run it [12:04] Ubit Umarov: so.. duhhh a bit stuck for now :) [12:05] Andrew Hellershanks: ok, Jamie. Thanks for coming. See you again next week. [12:05] Ubit Umarov: 2 late :) [12:05] Selby.Evans @grid.kitely.com:8002: bye all [12:05] Andrew Hellershanks: ok, Selby. See you next week. [12:05] Ubit Umarov: have fun sleby [12:05] Vincent.Sylvester @hg.zetaworlds.com:8002: Revert back to working version and then go step by step, time consuming but probably simplest way [12:05] Ubit Umarov: nad sleby [12:06] Ubit Umarov: dannng [12:06] Ubit Umarov: and selby [12:06] Andrew Hellershanks: We are just past the hour mark. If there aren't any last minute items for today then we can wrap up todays meeting. [12:06] Ubit Umarov: yeah i have rl waiting also [12:07] Andrew Hellershanks: ok. In that case I will call this meeting to a close. I also have a couple of things I need to get back to. [12:07] Ubit Umarov: have anice week ppl [12:07] Andrew Hellershanks: Thank you all for coming. See you again next week.