Réunion du 24-05-2022
Source : http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2022-05-24
Introduction
[11:09] Ubit Umarov : ouais c'est mieux d'avoir une maison plus petite.
[11:09] Andrew Hellershanks : Bonjour, tout le monde.
[11:09] Ubit Umarov : mais il est là.
[11:09] Ubit Umarov : donc il est possible qu'un plus grand bâtiment soit toujours ok.
[11:10] Vincent.Sylvester @hg.zetaworlds.com:8002 : Je m'en tiendrai à lancer des véhicules motorisés la prochaine fois.
[11:10] Andrew Hellershanks : J'ai eu des difficultés techniques mineures aujourd'hui. Une mise à jour récente du pilote de la carte vidéo m'a obligé à redémarrer mon ordinateur avant que Firestorm ne fonctionne.
[11:10] Ubit Umarov : certains blocs moteurs sont difficiles.
[11:10] Ubit Umarov : Selby est en train de se reloguer.
[11:10] Ubit Umarov : il y avait des problèmes avec les animations.
[11:11] Andrew Hellershanks : Bonjour, Selby. wb
[11:11] Ubit Umarov: mieux selby.Evans ?
[11:11] Ubit Umarov : je vois que tu es assis maintenant.
[11:12] Selby.Evans @grid.kitely.com:8002 : bonjour à nouveau -- réparé
[11:12] Ubit Umarov: bien
[11:12] Ubit Umarov : donc andrew... lance le metting quand c'est prêt.
[11:12] Ubit Umarov : oops log déjà activé
[11:12] Ubit Umarov : :)
[11:13] Andrew Hellershanks: :)
Lancement de rayon
[11:13] Andrew Hellershanks : Kayaker, as-tu remarqué qu'Ubit a fait un changement qui pourrait avoir résolu ton problème de ray casting ?
[11:13] Ubit Umarov : non, pas du tout.
[11:14] Ubit Umarov : peut être qu'il l'a amélioré.
[11:14] Kayaker Magic : Awww, tu es en train de briser tous mes espoirs....
[11:15] Ubit Umarov : Tampa a fait ce patch en inversant une partie de mon code... ce n'était pas une bonne inversion, mais il a remarqué une amélioration.
[11:15] Ubit Umarov : en fait, j'avais mal placé un appel à sqrt math.
[11:15] Ubit Umarov : coff coff coff
[11:15] Vincent.Sylvester @hg.zetaworlds.com:8002 : Je suis en train de compiler un binaire pour continuer les tests et voir ce que je peux trouver d'autre.
[11:16] Ubit Umarov : pour une longueur de rayon > taille de région, cela a un impact sur les performances, bien sûr.
[11:16] Vincent.Sylvester @hg.zetaworlds.com:8002 : Le commit original n'incluait pas de sqrt lors de l'élévation au carré de la distance, ce qui a déclenché une erreur de traitement quelque part.
[11:16] Ubit Umarov : le problème majeur se situe au niveau de la bibliothèque de physique.
[11:16] Ubit Umarov : comme je l'ai dit, la charge du rayon s'échelonne avec le carré de la longueur du rayon.
[11:17] Ubit Umarov : c'est juste mauvais... et peut être linéaire avec lui
[11:17] Ubit Umarov : il faut juste un meilleur algorithme de balayage.
[11:18] Ubit Umarov : ma perte du sqrt rendait les choses encore pires, bien sûr :)
[11:18] Ubit Umarov : si vous avez demandé 10m, si vous avez fait 100 ... oops ;)
[11:19] Vincent.Sylvester @hg.zetaworlds.com:8002 : Ouais, j'ai déclenché une sorte de boucle, je ne sais pas où, mais c'était assez mauvais, bien que je n'aie pas eu de plantage, juste des échecs de heartbeat.
[11:19] Ubit Umarov : non, pas une boucle.
[11:20] Ubit Umarov : le code absurde fait des contrôles de rayon sur une boîte 100x100 au lieu de 10x10.
[11:20] Ubit Umarov : ( on x,y)
[11:20] Ubit Umarov : comme je l'ai dit, je veux changer cela en une recherche linéaire... mais cela prendra plus de temps.
[11:21] Ubit Umarov : je veux dire une recherche sur une ligne, pas sur une case.
[11:21] Ubit Umarov : c'est un changement qui vient d'être fait aujourd'hui.
[11:21] Ubit Umarov : toujours à vérifier bien sûr !
[11:22] Ubit Umarov : d'autres changements ont été effectués la semaine dernière, laissez-moi voir.
Autres changements de la semaine
[11:22] Vincent.Sylvester @hg.zetaworlds.com:8002 : Don de code pour les paramètres du projecteur
[11:22] Ubit Umarov : oui
[11:23] Ubit Umarov : j'ai accepté un patch de lickx ? ??
[11:23] Vincent.Sylvester @hg.zetaworlds.com:8002 : Lotek
[11:23] Andrew Hellershanks : Ce doit être celui qui est lié à PRIM_PROJECTOR.
[11:23] Vincent.Sylvester @hg.zetaworlds.com:8002 : Yep
[11:24] Ubit Umarov : pour supporter PRIM_PROJECTOR sur les fonctions LlSetPrimitiveParams*().
[11:25] Ubit Umarov : également un patch de vicent supportant les *Get*.
[11:25] Ubit Umarov : le dernier a quelques problèmes, donc considérons-le comme une acceptation temporaire.
[11:25] Andrew Hellershanks : Je vais devoir regarder cela. Je ne me souviens pas de ce que c'est censé faire.
[11:25] Ubit Umarov : SL ne l'a pas
[11:26] Ubit Umarov : le truc est de lire les ID des textures.
[11:26] Ubit Umarov : sl bloque cela
[11:26] Vincent.Sylvester @hg.zetaworlds.com:8002 : SL a enlevé la partie Get pour des "raisons de sécurité" ouais, ce n'est pas vraiment une option dans OpenSim.
[11:26] Vincent.Sylvester @hg.zetaworlds.com:8002 : Vérifier llGetTexture pour savoir si le propriétaire de la prim a cette texture dans son inventaire est une vérification très lourde à faire.
[11:27] Ubit Umarov : dans la plupart des cas lsl ne peut afficher une texture (texture, material maps, etc) que si le propriétaire du script a tous les droits sur la prim.
[11:27] Andrew Hellershanks : S'il y avait une préoccupation à ce sujet, une option aurait été de faire de la fonction get une fonction OSSL pour avoir une autorisation pour cela.
[11:27] Ubit Umarov : dans le cas du projet, il ne supporte pas get.
[11:27] Ubit Umarov : listé sur le wiki mais ne fonctionne pas. [11:28] Vincent.Sylvester @hg.zetaworlds.com:8002 : llGetTexture n'a déjà pas de contrôles de sécurité, SL vérifie des choses là-dessus, donc je me suis dit que ça n'avait pas d'importance.
[11:28] Vincent.Sylvester @hg.zetaworlds.com:8002 : Il ne s'appelle pas ClosedSim, n'est-ce pas ?
[11:28] Ubit Umarov : pour l'instant nous ne limitons pas la lecture des UUIDs des textures sur LSL/OSSL.
[11:28] Ubit Umarov : donc pour l'instant je l'accepte sans vérification.
[11:29] Ubit Umarov : c'est quelque chose que nous devrons revoir un jour...
[11:29] Andrew Hellershanks hoche la tête
[11:29] Ubit Umarov : mais cacher les UUIDs comme le fait SL n'est pas exactement une fonction de sécurité.
[11:29] Ubit Umarov : oh, en rapport avec cela, j'ai restreint un peu l'ossl.
[11:30] Ubit Umarov : nos paramètres osSet/get sur les autres prims.
[11:30] Ubit Umarov : maintenant non seulement le propriétaire doit être le même, mais il doit aussi avoir les droits de modification.
[11:30] Ubit Umarov : pas besoin sur la même prim...
[11:31] Ubit Umarov : hmm en fait nous devrions changer cela pour le même linkset
[11:31] Vincent.Sylvester @hg.zetaworlds.com:8002 : Je suis heureux d'annoncer que je viens de tester raycast à nouveau sur master et qu'il indique à nouveau avoir trouvé la prim et que le temps est de 0.051200, donc ce commit a définitivement aidé à résoudre le pire des problèmes :)
[11:32] Ubit Umarov : a aussi changé le propriétaire de la prim utilisée, pour être la prim root, celle qui est vraiment valide.
[11:32] Andrew Hellershanks : Ubit, ça sonne bien.
[11:32] Andrew Hellershanks : Vincent, merci pour le rapport de test.
[11:32] Ubit Umarov : notre code devrait fixer le propriétaire sur toutes les prims d'un linkset... mais... devrait... pas sûr que ce soit le cas...
[11:33] Ubit Umarov : de toute façon, le seul valide est celui de la prim racine... donc toutes nos vérifications devraient se faire à ce niveau...
[11:33] Ubit Umarov : au moins changé en cela à cet endroit :)
[11:33] Andrew Hellershanks hoche la tête
[11:33] Ubit Umarov : en fait le propriétaire ne devrait être un paramètre d'une partie.
[11:34] Ubit Umarov : nous avons juste un linkset ou sceneobjectgroup dans notre code.
[11:34] Ubit Umarov : mais les viewers penseront aussi que les pièces ont des propriétaires... donc... des détails.
[11:35] Vincent.Sylvester @hg.zetaworlds.com:8002 : J'ai aussi supprimé quelques casts de l'api lsl qui n'étaient pas nécessaires, ce qui a permis de gagner quelques nanosecondes :)
[11:36] Andrew Hellershanks: :)
[11:36] Ubit Umarov : oups le coup de téléphone m'a aussi fait oublier de mentionner sur la mantis le changement de code pour les rayons.
[11:36] Ubit Umarov : je me demande pourquoi fs décide de chuchoter :)
[11:37] Ubit Umarov : juste ns sur compiler, le compilateur n'est pas si stupide :) (pas de version chuchotée)
[11:37] Vincent.Sylvester @hg.zetaworlds.com:8002 : Il faut déjà 20 minutes pour compiler, donc tout aide, surtout les tests lents et le fait que Jenkins soit bizarre.
[11:37] Andrew Hellershanks : Ubit, touche shift enfoncée lorsque tu appuies sur entrée, IIRC.
[11:37] Andrew Hellershanks murmure : test...
Les inventaires
[11:38] Ubit Umarov: meanwhile on Firestorm Jira etc some more talks abotu that new inventory checks [11:38] Andrew Hellershanks: Yup. that's it. [11:38] Ubit Umarov: firestorm mostly just turned all them just warnings on opensim [11:38] Ubit Umarov: but some more talk went on abotu this or that detail [11:39] Ubit Umarov: no idea abotu other viewers [11:39] Ubit Umarov: think alchemy had just fixed that not merging the ll code for opensim [11:40] Ubit Umarov: hmm there was some other thing on viewers? [11:40] Vincent.Sylvester @hg.zetaworlds.com:8002: I like the idea of fixing inventories up a bit, helps with IARs actually going through and not failing half way through, so some good stuff in there, just was a bit strict for OpenSim modular design [11:40] Vincent.Sylvester @hg.zetaworlds.com:8002: I did mention to FS and Alchemy that we had that RSS feed for grid status now [11:41] Vincent.Sylvester @hg.zetaworlds.com:8002: They seem to want to implement that at some point [11:41] Ubit Umarov: wel users should merge dupicated system folders [11:41] Ubit Umarov: outch typos [11:42] Ubit Umarov: grids can try some clean up on dbs [11:42] Ubit Umarov: but as soon they load a IAR they can get "bad things" again [11:42] Kayaker Magic: My alt also reports that the llCastRay changes this morning fixed the problem [11:42] Ubit Umarov: as i said, i have no planes to remove the IAR duplication of folders [11:43] Ubit Umarov: unless the option -merge is added [11:43] Ubit Umarov: because that does help users to proper merge them [11:43] Vincent.Sylvester @hg.zetaworlds.com:8002: I think that is actually a good approach, because in some cases you want to effectively do a replace of inventory rather than merging anyways [11:43] Ubit Umarov: ...or allows users to merge them... [11:44] Ubit Umarov: yeah, that also [11:44] Vincent.Sylvester @hg.zetaworlds.com:8002: DB cleanup is something grid operators need to do anyways given the garbage that can collect if not careful [11:44] Ubit Umarov: can't have a automatic solution for all cases [11:45] Ubit Umarov: and to be real that is not and should never be a issue for viewers [11:45] Vincent.Sylvester @hg.zetaworlds.com:8002: Most cases users can fix themselves even, only needs admin intervention in really broken cases [11:45] Ubit Umarov: as i said the only case then want a clean system folder is when deciding where to put a new created item [11:46] Ubit Umarov: but that they can put on the just the frst they find that fits criterias like folder type [11:47] Ubit Umarov: so several copies with same name and type, is not a breaking issue [11:47] Ubit Umarov: just ugly and confusing to have [11:47] Vincent.Sylvester @hg.zetaworlds.com:8002: We also had some fun with mono project this week, Ubit went full berserk-mode :D [11:48] Ubit Umarov: but most users have a grandson that will he happy to help cleanup inventory [11:48] Ubit Umarov: ahh yeah ... mono.. jezz [11:49] Ubit Umarov: mono made no new release since Feb last year [11:49] Ubit Umarov: then last week, just decided to make one.. [11:49] Ubit Umarov: and made a big mess on the packages for linux [11:49] Ubit Umarov: so they where useless [11:50] Ubit Umarov: we even did got a mantis on that [11:50] Ubit Umarov: several ppl, inc vicent, had report it, telling the exact issue etc [11:50] Ubit Umarov: and they ignored it [11:50] Vincent.Sylvester @hg.zetaworlds.com:8002: Basically forgot to bump the msbuild version in the latest release so mismatched that with the supporting libraries making compile fail, could be fixed by adding their preview repo which was the same version, but they did now fix their stable repo as well [11:51] Ubit Umarov: i went rant to discord, ofc :p [11:51] Vincent.Sylvester @hg.zetaworlds.com:8002: That got their attention [11:51] Ubit Umarov: yeah.. it took that.. [11:51] Kayaker Magic: Is it safe to update to latest mono yet? [11:51] Ubit Umarov: so yesterday they fixed the packages.. ( at least some) [11:52] Ubit Umarov: then forgot the page, still only telling about the older versions [11:52] Ubit Umarov: me at discord again.. [11:52] Ubit Umarov: and they fixed the page [11:52] Ubit Umarov: it now does tell thye did release .. 179 [11:53] Ubit Umarov: ofc they still forgot the release notes.. [11:53] Andrew Hellershanks: :) [11:53] Vincent.Sylvester @hg.zetaworlds.com:8002: Which yes so far seems to work fine, slightly faster compiles according to jenkins [11:53] Ubit Umarov: se me [11:53] Ubit Umarov: Ubit — Yesterday at 6:00 PM yeah looks better, now the release notes on the site? [6:00 PM] https://www.mono-project.com/docs/about-mono/releases/ [11:53] Ubit Umarov: and ok jthey did read me... [11:53] Ubit Umarov: so: [11:54] Ubit Umarov: Ubit — Yesterday at 6:58 PM cool see release notes for Mono 6.12.0.174 (22 Mar 2022) [6:58 PM] can we now have the ones for Mono 6.12.0.179 (22 Mar 2022) ? [11:54] Ubit Umarov: they ignore me :p [11:55] Ubit Umarov: they did release 6.12.0.179 but only show release notes for a 174 they never did release [11:55] Ubit Umarov: well [11:55] Ubit Umarov: this is very sad [11:55] Ubit Umarov: just shows how current MS totally ignores mono [11:55] Vincent.Sylvester @hg.zetaworlds.com:8002: Ubit "No Chill" Umarov heh, though I agree that is really bad way to handle release, they get paid for this... [11:56] Ubit Umarov: for them mono is just dead.. keeping it just to "look cool.." [11:57] Vincent.Sylvester @hg.zetaworlds.com:8002: Keeping the illusion of caring for open source, while they just use it on their cloud to make bank, soon as it makes them no more money it will get dropped like hot potato [11:58] Ubit Umarov: they also have commits like mine!!! ie like "fix some bugs" [11:58] Ubit Umarov: lol [11:58] Andrew Hellershanks: hehe [11:58] Andrew Hellershanks: Always a very informative commit message. :P [11:58] Ubit Umarov: they have a big list of pull requests they mostly ignore [11:59] Ubit Umarov: and a ton of issue reports there getting dust [11:59] Vincent.Sylvester @hg.zetaworlds.com:8002: Over 2k issues on github no wonder important issues get lost [11:59] Vincent.Sylvester @hg.zetaworlds.com:8002: Why you think I been working to clean mantis, you need to know what's wrong if you want to fix things [11:59] Ubit Umarov: i almost could read how bored the guy was, going to do the fix :p [12:00] Vincent.Sylvester @hg.zetaworlds.com:8002: heh [12:00] Ubit Umarov: ffor them dotnet is the only golden eggs chicken they care about [12:01] Ubit Umarov: fun think on release 122 last year i also had to tell then on discord that there where no release notes [12:01] Vincent.Sylvester @hg.zetaworlds.com:8002: Thankfully it still is open source so if Xamarin continues to screw things up there is the small chance some competent people fork it [12:02] Ubit Umarov: samarin is gone [12:02] Ubit Umarov: mono is now supposed to be owned by a foundation [12:02] Ubit Umarov: that is mostly owned by ms [12:03] Ubit Umarov: well it was a huge mistake to make opensim on c# [12:03] Ubit Umarov: but no point keep talking about that.. [12:04] Vincent.Sylvester @hg.zetaworlds.com:8002: Not as big a mistake as making it on c++ lol [12:04] Vincent.Sylvester @hg.zetaworlds.com:8002: But yeah now we deal with it best we can [12:04] Ubit Umarov: nah c++ is terrible, but everything else is trash :p [12:04] Vincent.Sylvester @hg.zetaworlds.com:8002: Lots of elbow grease and swearing [12:04] Andrew Hellershanks: :) [12:05] Ubit Umarov: ofc c would be better.. but lets not go that far :p [12:06] Ubit Umarov: c++ is even worse than c, since allows for a ton of diferent solutions to pile up [12:06] Ubit Umarov: ohh and insane abuse of macros [12:06] Vincent.Sylvester @hg.zetaworlds.com:8002: Porting OpenSim to C and before you know it you find yourself writing kernel patches... [12:06] Ubit Umarov: C# did had a more clean language definition [12:07] Ubit Umarov: a lot nicer to read and write than c++ [12:07] Ubit Umarov: problem it the stupid managed part [12:07] Ubit Umarov: total failed GC [12:08] Vincent.Sylvester @hg.zetaworlds.com:8002: c++ put me off coding when I first started to learn it for school, it was so backwards and weird for simple tasks, c# is beginner friendly [12:08] Ubit Umarov: JIT mess etc [12:08] Andrew Hellershanks: I've also found that any project written in C++ code takes a lot longer than any C based project. [12:08] Ubit Umarov: i once did read a book abotu c++ [12:08] Vincent.Sylvester @hg.zetaworlds.com:8002: And most c++ is like you need boost and cmake and then this and that dependency, total mess to get working at time, case in point, viewers [12:08] Andrew Hellershanks: I might have a book around here on C++ [12:09] Ubit Umarov: sent it the the trash when i did reach "friend classes" [12:09] Ubit Umarov: what a **** bahh [12:09] Ubit Umarov: that was a total negation of all told before about objects lol [12:09] Vincent.Sylvester @hg.zetaworlds.com:8002: Not even the good kind of friends with benefits, friends with demands more like [12:10] Ubit Umarov: lets not go that path :P [12:10] Ubit Umarov: so... what news do you have about opensim? [12:10] Ubit Umarov: :) [12:12] Vincent.Sylvester @hg.zetaworlds.com:8002: Was testing the whole thing with the objects not deleting properly and thought I could simulate that by making a lot of prims, so wrote my first auto replication script to rezz tons of prims, promptly forgot it was on state entry and when testing yesterday it made over 3 million prims [12:12] Vincent.Sylvester @hg.zetaworlds.com:8002: Finally used all the memory on my computer though [12:13] Vincent.Sylvester @hg.zetaworlds.com:8002: So I now know that all sticks work properly lol [12:13] Ubit Umarov: lol [12:13] Andrew Hellershanks: lol [12:14] Kayaker Magic: Self replicating objects is what attracted me to SL back in the day. But it is hard to prevent it from running away from you.... [12:15] Vincent.Sylvester @hg.zetaworlds.com:8002: I set it to only make 10k prims, but then I kept rebuilding and testing so restarted like 200 times or more and it kept making more and more [12:18] Andrew Hellershanks: nebadon has some scripts that make some objects which involved a huge number of prims. [12:19] Andrew Hellershanks: The Klein bottle and a Sierpinski triangle. [12:19] Andrew Hellershanks: We are getting close to half past the hour. Any last minute topics for today? [12:20] Andrew Hellershanks: I don't see anyone typing so I will take that as a no. [12:20] Andrew Hellershanks: ON that basis I will close todays meeting. Thank you all for coming. See you again next week. [12:20] Selby.Evans @grid.kitely.com:8002: Bye all [12:20] Andrew Hellershanks: Bye, Selby [12:21] Jamie.Jordan @grid.kitely.com:8002: thanks everybody