Réunion du 10-05-2022

Source : http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2022-05-10

Introduction

[11:02] Ubit.Umarov : j'ai changé les sources et mis à jour bin/ en conséquence.

[11:02] Andrew Hellershanks : Bonjour à tous.

[11:02] Michael.Christopher @cc.opensimulator.org:8002 : bonjour !

[11:02] Motoko.Karu @login.digiworldz.com:8002 : Hihi

[11:03] Michael.Christopher @cc.opensimulator.org:8002 : Bonjour Motoko, ravi de te rencontrer

[11:03] Misterblue Waves : Bonjour à tous.

[11:03] Ubit.Umarov : la dernière tentative d'exclure les dossiers objP n'a pas très bien fonctionné.

[11:03] Ubit.Umarov : ils ne sont pas ignorés.

[11:03] Andrew Hellershanks : Ravi de te revoir, Misterblue.

Modifications de code cette semaine

[11:05] Vincent Sylvester : Quelques ajustements mineurs dans le code cette semaine,

• refonte de certains assets de la bibliothèque,
• travail sur le support des permissions appropriées bien que les viewers aient encore des problèmes avec les permissions d'Export.
• Ajout d'un support pour cacher le titre de l'avatar.
• Ajustements aux listes, pour gagner quelques nanosecondes héhé.

Problème des inventaires : les types des dossiers

• Script de contrôle d'inventaire  : Avant 2022, les viewers acceptaient toute forme d'inventaire en dehors des configurations complètement mal formées. Cela a maintenant changé avec un nouveau code ajouté aux viewers conçu pour valider si la structure de l'inventaire est correctement configurée.

[11:05] Vincent Sylvester : Les choses concernant le code de vérification de l'inventaire dans les viewers sont toujours en cours de discussion, mais il est clair qu'il n'y a pas de solution facile, les administrateurs de grille devront faire un travail de fond pour corriger les inventaires.

[11:06] Ubit.Umarov : Je n'ai toujours rien fait à ce sujet.

[11:06] Vincent Sylvester : Je ne pense pas que nous devrions.

[11:07] Vincent Sylvester : Nettoyer la base de données et s'assurer que le désordre qui s'y trouve est réglé fait partie du rôle d'un administrateur de grille. Laisser OpenSim faire le ménage lui-même, laisser n'importe quel logiciel le faire, est une mauvaise décision.

[11:07] Vincent Sylvester : La confiance c'est bien, la vérification c'est mieux.

[11:07] Ubit.Umarov : ce sont principalement des règles de LL.

[11:07] Ubit.Umarov : cela n'a pas eu d'impact sur le travail actuel.

[11:07] Andrew Hellershanks : Bonjour, Selby

[11:08] Ubit.Umarov : et très difficile à changer.

[11:08] Ubit.Umarov : au moins depuis la correction de 2015.

[11:08] Ubit.Umarov : avec les patchs sur les types de dossiers faits par Drake darconis sur libopenmetaverse et appliqués à opensim par diva.

[11:09] Ubit.Umarov : même ainsi, les choses plus anciennes peuvent encore avoir de mauvais types.

[11:09] Vincent Sylvester : J'ai dû corriger les inventaires de 3 utilisateurs jusqu'à présent, dont l'un des miens, c'est assez simple de changer les types de dossiers en dossiers normaux, c'est assez simple à faire si vous pouvez trouver le bon dossier.

[11:09] Ubit.Umarov : spécialement les choses sur les IARs.

[11:09] Ubit.Umarov : mais c'est encore pire...

[11:09] Ubit.Umarov : par défaut, l'IAR ne fusionne pas les dossiers avec le même nom.

[11:10] Ubit.Umarov : donc, il va dupliquer tous les dossiers système au chargement.

[11:10] Ubit.Umarov : comme 2, 3 etc. Dossiers des notecards, etc.

[11:10] Vincent Sylvester : En ce moment, je charge habituellement les IAR dans des sous-dossiers en laissant aux utilisateurs le soin d'extraire ce dont ils ont besoin dans les dossiers d'inventaire.

[11:10] Ubit.Umarov : c'est une violation de ces "nouvelles règles".

[11:10] Ubit.Umarov : ouais mais suivant les spec on doit pouvoir charger aussi sur /.

[11:11] Ubit.Umarov : c'est donc une autre raison pour laquelle les nouveaux firestorm (fs) (et bientôt d'autres viewers) refusent de charger.

[11:11] Ubit.Umarov : et ce n'est pas facile de repérer tous les cas et de les corriger.

[11:12] Ubit.Umarov : aussi pas agréable pour les très grandes bases de données.

[11:12] Ubit.Umarov : L'idée de Beq est de faire en sorte que FS fasse un avertissement sans échec.

[11:12] Ubit.Umarov : et cela semble la meilleure solution.

[11:13] Vincent Sylvester : Je veux dire que la conception de la base de données de l'inventaire ainsi que quelques autres tables est abyssale c'est vrai, il y a maintenant d'autres concepts de base de données qui fonctionneraient mieux pour le type de données que l'inventaire gère, mais reconstruire l'interface de la base de données et exiger un autre composant logiciel ajoute aussi de la complexité et des risques d'échec.

[11:13] Ubit.Umarov : certains pourraient même argumenter que d'avoir 5 dossiers nommés Notecards tous de type 7 sous My inventory est une fonctionnalité normale d'opensim (et c'est le cas jusqu'à présent).

[11:13] Vincent Sylvester : Je ne pense vraiment pas que les viewers devraient autoriser les mauvais inventaires à l'avenir, parce que s'il y a une chose que je sais, c'est que si rien n'est jamais vérifié, ça va juste s’aggraver progressivement.

[11:14] Ubit.Umarov : ce ne sont pas vraiment de mauvais inventaires.

[11:14] Vincent Sylvester : C'est vraiment pénible quand on veut construire quelque chose et qu'on se fie aux spécifications pour découvrir qu'elles ne sont pas appliquées et qu'il faut gérer des tonnes de cas particuliers.

[11:15] Ubit.Umarov : les mauvais inventaires sont ceux qui ont perdu leur parent, les liens pointant vers des choses manquantes, etc...

[11:16] Ubit.Umarov : bon, la plupart du code des inventaires a été écrit quand les devs d'opensim avaient la règle de ne pas voir les sources des viewers.

[11:16] Vincent Sylvester : Eh bien, concentrons-nous d'abord sur les erreurs fatales, c'est-à-dire principalement les mauvais types et les dossiers racines, qui ne sont pas si difficiles à corriger.

[11:16] Ubit.Umarov chuchote : même libopenmetaverse était mal en point.

[11:16] Ubit.Umarov : comme je l'ai dit, c'est Drake qui a corrigé libopenmetaverse avec des types de dossiers corrects.

[11:17] Ubit.Umarov : jusqu'alors, libomv supposait aussi que foldertype était la même chose que assetType.

[11:17] Vincent Sylvester : Je suis en train d'écrire un petit script pour explorer mes propres tables à la recherche de mauvaises données afin que je puisse les corriger avant d'être submergé de tickets, je mettrai un lien vers ce script sur le wiki une fois que je l'aurai terminé et testé.[1]

[11:18] Ubit.Umarov : les viewers doivent transformer ces contrôles en simples avertissements.

[11:18] Ubit.Umarov : Il y a deux cas qui ne suivent pas ces règles à corriger.

[11:18] Ubit.Umarov : pour commencer, nous devons détruire le code IAR actuel.

[11:18] Ubit.Umarov : tous cassés selon ces règles.

[11:19] Vincent Sylvester : Les changer en "non fatal" signifiera simplement que tout ce code ne sert à rien car les avertissements seront simplement ignorés.

[11:19] Ubit.Umarov : une autre alternative est de tuer la plupart des grilles opensim.

[11:20] Vincent Sylvester : Non, réparer les erreurs fatales n'est pas aussi difficile qu'il n'y paraît.

[11:20] Vincent Sylvester : Trivial

[11:20] Ubit.Umarov : y compris la version actuelle 0.9,2,2

[11:20] Ubit.Umarov : cela va dupliquer les dossiers du système.

[11:20] Ubit.Umarov : sur les chargements d'IAR

[11:20] Ubit.Umarov : quand même

[11:20] Ubit.Umarov : non, ce n'est pas si facile.

[11:21] Ubit.Umarov : il faut tout reparenter sur un dossier dupliqué, etc etc

[11:21] Ubit.Umarov : une chose effrayante sur une grille avec 1M d'entrées ou beaucoup plus...

[11:21] Ubit.Umarov: et pour AUCUNE raison

[11:22] Ubit.Umarov : les choses fonctionnaient bien et fonctionnent toujours.

[11:22] Ubit.Umarov : et un tel undump est quelque chose que les utilisateurs peuvent faire à la main dans leur temps libre.

[11:22] Ubit.Umarov : beaucoup plus sûr (on l'espère) que les scripts sql aveugles.

[11:23] Vincent Sylvester : Je ne suis pas habituellement du côté du viewer, mais dans ce cas je pense que s'assurer que l'inventaire est correct est fondé. Quand au code d'IAR désordonné d'OpenSim, c'est quelque chose qui techniquement n'est pas cassé, juste pas très bien reconnu peut-être, mais ce n'était pas non plus prévu.

[11:23] Ubit.Umarov: bien sûr, le type 9 de My inventory devrait être rectifié avec une requête SLQ

[11:23] Ubit.Umarov: ou le type 8 de My suitcase

[11:24] Vincent Sylvester : C'est un système de sauvegarde débile fait plutôt pour fonctionner efficacement comme un remplacement, fusionner les choses et ne pas écraser l'inventaire entier semble ne pas avoir été vraiment pensé.

[11:24] Ubit.Umarov : l'option "non fusion" est simplement cassée.

[11:24] Ubit.Umarov : au moins sur les dossiers du système directement sous My inventory

[11:24] Ubit.Umarov : mais bon, ces règles sont, encore une fois, celles de LL.

[11:24] Vincent Sylvester : Il y a une solution facile à cela, interdire le chargement de l'inventaire dans / et forcer tous les dossiers à être de type -1.

[11:25] Ubit.Umarov : l'idée que mon inventaire doit avoir EXACTEMENT des dossiers précis n'est pas cool.

[11:25] Ubit.Umarov : les viewers pour opensim ont besoin de plus de flexibilité dans la limite du raisonnable.

[11:26] Vincent Sylvester : Je suis d'accord qu'OpenSim devrait offrir de la flexibilité, mais si nous voulons changer la spécification des choses, nous devons documenter cela pour que les viewers aient une référence sur laquelle se baser.

[11:26] Vincent Sylvester : Le wiki ne montre que la spécification LL pour le moment.

[11:26] Ubit.Umarov : donc transformer cette nouvelle validation en un simple avertissement est un pas dans la bonne direction.

[11:26] Ubit.Umarov : avec ces logs, nous pouvons aussi avoir une meilleure idée de ce qui existe.

[11:27] Ubit.Umarov : en supposant que certains utilisateurs qui ont des problèmes partagent les logs :)

[11:27] Vincent Sylvester: No one will be sending in their logs or complain to a grid operator about warnings, especially ones in the logs almost no one opens anyways [11:27] Ubit.Umarov: well some will, if you ask [11:27] Ubit.Umarov: nd the is enough [11:28] Vincent Sylvester: At that point might as well just check the table yourself using a script that applies the same validation rules [11:28] Ubit.Umarov: well another case in last days is the new viewers option to mute collision sounds [11:28] Ubit.Umarov: sadly viewers are just broken on that [11:29] Vincent Sylvester: There still remain quite a few things viewers are broken on, though lately the renewed interest in viewer development is bringing them to light one by one [11:29] Ubit.Umarov: thing is that they just compare a sound UUID with list of known collision sounds, and only mute those [11:29] Vincent Sylvester: Just takes time and effort to find and fix them all [11:29] Ubit.Umarov: ofc opensim has diferent UUIDs, becasue there was no need to have he same [11:30] Ubit.Umarov: and so that broken viewer option does not work on opensim [11:30] Ubit.Umarov: and it is broken, bc who made it forgot about llCollsionSound [11:30] Ubit.Umarov: that can change the sound to any other sound [11:30] Vincent Sylvester: I was about to ask about that [11:30] Vincent Sylvester: Do collision sounds not send to the viewer with a special type? [11:31] Ubit.Umarov: so just today i made a new code, viewers may use or not [11:31] Ubit.Umarov: added use of bit 7 of the Flags viewers send on agent Updates [11:32] Ubit.Umarov: if that is set, the region will not send collsion sounds to that avatar [11:32] Ubit.Umarov: that code is running on region ubittest here [11:32] Ubit.Umarov: and told abotu it to FS and alchemy [11:33] Ubit.Umarov: viewer code is simple, so i hope all will accept and support [11:33] Ubit.Umarov: region does Know abotu llCollisionSound, ofc [11:33] Ubit.Umarov: this flag ( or similar) is the only way to do it... not more silly "magic UUIDs" [11:34] Ubit.Umarov: well as we type, our site is DOWN [11:34] Ubit.Umarov: and no idea when even if it will return [11:34] Vincent Sylvester: If a sound used as collision sound had a special type muting that would be easy to do [11:34] Michael.Christopher @cc.opensimulator.org:8002: ?? no idea IF? [11:34] Andrew Hellershanks: Yes.Not sure when it went offline. [11:34] Arielle Popstar: how long has it been down? [11:34] Arielle Popstar: who runs it? [11:34] Ubit.Umarov: sometime in last ours [11:35] Ubit.Umarov: hours [11:35] Ubit.Umarov: went down during morning to me [11:35] Arielle Popstar: noticed other times its been down too ovewr the past month [11:35] Ubit.Umarov: i did commit sounds changes last night [11:35] Ubit.Umarov: today was doing he commit of the new flag code and poff [11:35] Ubit.Umarov: machine totally unreachable [11:36] Ubit.Umarov: including the Virtual machine supervisor [11:36] Andrew Hellershanks: Arielle, This outage is not the same as the ones we have expierience in the past. [11:36] Arielle Popstar: oh [11:36] Ubit.Umarov: i did warn the only person than can talk with the housing ppl [11:36] Arielle Popstar: server or hosting troubles? [11:36] Ubit.Umarov: need to wait now [11:37] Ubit.Umarov: seem the machine was intencionaly blocked [11:37] Ubit.Umarov: could had been like a stronf DDOS [11:37] Michael.Christopher @cc.opensimulator.org:8002: ouch [11:37] Ubit.Umarov: some housing ppl will kisck you out on such cases [11:37] Arielle Popstar: we have offline backups? [11:37] Ubit.Umarov: but no real information.. we need to wait [11:37] Misterblue Waves: or it's running such an old version of everything it was compromised [11:38] Misterblue Waves: these days you have to keep your patches up to date [11:38] Ubit.Umarov: meanwhie we have 2 git repos im keeping updated by hand [11:38] Ubit.Umarov: https://bitbucket.org/opensimulator/ [11:38] Vincent Sylvester: The entire machine needs rebuilding at this point, but it seems that keeps being postponed [11:38] Ubit.Umarov: here i keep all our repos [11:39] Ubit.Umarov: i created that bitbucket also on a ocasion our box did vanish [11:39] Ubit.Umarov: that time bc was unpaid :p [11:39] Misterblue Waves: I do all my work on Github so my preference is to move there but others are so hot on that idea [11:39] Ubit.Umarov: we lao have the gthub repo [11:40] Ubit.Umarov: that i need to keep updated by hand also [11:40] Misterblue Waves: I was looking into adding build Actions there [11:40] Misterblue Waves: would need to figure out how to import Mantis and the Wiki [11:40] Ubit.Umarov: bc the old machine ssh is obsolete and now refused by github [11:40] Ubit.Umarov: https://github.com/opensim [11:41] Vincent Sylvester: The box is so ancient, I'm not gonna say anything, but it is [11:41] Ubit.Umarov: so in terms of code repo, we are ok [11:41] Ubit.Umarov: my code from today is on those 2 [11:41] Misterblue Waves thinks the ssh problem is because newer sites don't accept TLS v1 and the old machine doesn't have TLS v2 for ssh [11:41] Ubit.Umarov: everything else.. fingers crossed [11:42] Ubit.Umarov: yeha its openssl is not banned everywhere ( tls2 min ) [11:42] Vincent Sylvester: Melanie is the one in charge of the box, so far through workload probably hasn't had the time to rebuild it completely [11:42] Ubit.Umarov: is now.. [11:42] Vincent Sylvester: I offered help numerous times [11:42] Ubit.Umarov: yes she wants to do those updattes and is doing them since 2018 ?? [11:43] Arielle Popstar: is she still part of core? [11:43] Ubit.Umarov: meanwhile osgrid assets seems to have issues again [11:43] Ubit.Umarov: also waiting for her to fix [11:43] Andrew Hellershanks: Ubit, again? They have been having an awful lot of problems of late. [11:44] Ubit.Umarov @hg.osgrid.org:80: well now there is some vodoo [11:44] Ubit.Umarov @hg.osgrid.org:80: assets uploads work on some regions [11:44] Ubit.Umarov @hg.osgrid.org:80: and are lost if doen on others, like this or lbsa [11:44] Ubit.Umarov @hg.osgrid.org:80: we found that just sharing snapshots [11:45] Ubit.Umarov @hg.osgrid.org:80: well more code changes last week [11:45] Ubit.Umarov @hg.osgrid.org:80: i chnaged ubode [11:45] Arielle Popstar: are they osgrid only problems or other grids running newer code see the same? [11:46] Ubit.Umarov @hg.osgrid.org:80: opos ubode was the week before already :) [11:47] Ubit.Umarov @hg.osgrid.org:80: fixed a bug on reading ExportSupported option [11:47] Ubit.Umarov @hg.osgrid.org:80: it was read as string and should be as boolean [11:48] Ubit.Umarov @hg.osgrid.org:80: did retouch the collision sounds, making them similar lenght and removing some cliping they had [11:48] Ubit.Umarov @hg.osgrid.org:80: but small amplittude redution [11:48] Ubit.Umarov @hg.osgrid.org:80: minor thing [11:49] Ubit.Umarov @hg.osgrid.org:80: ann another option viewers had is hide title [11:49] Ubit.Umarov @hg.osgrid.org:80: i added suport for it [11:49] Ubit.Umarov @hg.osgrid.org:80: also today [11:50] Ubit.Umarov @hg.osgrid.org:80: that one does send a flag to region, in fact on same byte [11:50] Ubit.Umarov @hg.osgrid.org:80: and we wehre ignoring it [11:51] Arielle Popstar: you mean like the username? [11:51] Ubit.Umarov @hg.osgrid.org:80: group title above our heads [11:51] Ubit.Umarov @hg.osgrid.org:80: my "owner of nothing" [11:52] Arielle Popstar: did you ever allow usernames? [11:52] Ubit Umarov: preferences _ general has the option Hide Own group title [11:52] Ubit Umarov: usernames is what we have [11:52] Ubit Umarov: not the Display names [11:53] Ubit Umarov: and no . no plans to suport those.. we all hate them ;) [11:53] Andrew Hellershanks: :) [11:53] Arielle Popstar: I dont have usernames checked, only display names but i still see everyones name [11:53] Michael.Christopher @cc.opensimulator.org:8002 whispers: :)) [11:53] Ubit Umarov: there was a but on ubode i worked around [11:53] Ubit Umarov: even a bug! [11:54] Arielle Popstar: just you and andrew hate them but why should that matter [11:54] Andrew Hellershanks: We aren't the only two. [11:54] Arielle Popstar: only 2 i remember :) [11:54] Ubit Umarov: ofc can't go see what mantis 8995 is, but it was it :) [11:54] Michael.Christopher @cc.opensimulator.org:8002: only 2 that prefer usernames to display names? [11:54] Arielle Popstar: the other couldnt really care less [11:54] Arielle Popstar: that didnt want to allow the option [11:55] Vincent.Sylvester @hg.zetaworlds.com:8002: OpenSim is already a free for all in so many ways, give people the ability to change names at random and we get total anarchy [11:55] Arielle Popstar: eh? [11:55] Michael.Christopher @cc.opensimulator.org:8002: ok ty vincent, now I know what we are actually talking about [11:55] Ubit Umarov: LSL lists have a fun issue with null [11:56] Andrew Hellershanks: If you decide to have a display name that uses upside down backwards text how is someone to know how to type that so they can address a message to you? [11:56] Ubit Umarov: even on core c#code [11:56] Arielle Popstar: copy paste it [11:56] Ubit Umarov: comparing l == null calls a list to list compare [11:56] Andrew Hellershanks: If you are using a display name and someone has their viewer set to show only user names will you still recognize your own user name if someone uses it? [11:56] Michael.Christopher @cc.opensimulator.org:8002: makes sense now, i misread it the first time through :) [11:56] Ubit Umarov: that had somewhat slow code [11:56] Arielle Popstar: thats what i do in s/l [11:57] Ubit Umarov: in particula if running the region inside VS debug [11:57] Ubit Umarov: but slow as in very slow.. [11:57] Michael.Christopher @cc.opensimulator.org:8002: Andrew - While I would, I see the point - too many people wouldn't [11:57] Andrew Hellershanks: Copy it from where, Arielle? [11:57] Ubit Umarov: a changed that code [11:57] Ubit Umarov: so now it is slow as normal slow! [11:57] Arielle Popstar: but people rarely use local for direting chat to specific person. Most is done in IM [11:58] Arielle Popstar: profile [11:58] Arielle Popstar: or the radar list of names [11:58] Andrew Hellershanks: Profile shows the username. [11:58] Ubit Umarov: display names where the WORSE idea lindens had :p [11:58] Arielle Popstar: not in S/L [11:59] Ubit Umarov: ove the ones with just space [11:59] Ubit Umarov: we see Nothing [11:59] Arielle Popstar: they have a lot of bad ideas but Opensim still follws them for everything else [11:59] Andrew Hellershanks: I don't have any radar device. [11:59] Ubit Umarov: see a case we did not follow :p [11:59] Vincent.Sylvester @hg.zetaworlds.com:8002: Are we really having that debate again? Display names is a no, that's that [12:00] Michael.Christopher @cc.opensimulator.org:8002: Sorry, I've got to hop out a hair early, wife needs this computer for class about to start [12:00] Arielle Popstar: The people icon for nearby, friends, groups etc [12:00] Andrew Hellershanks: ok, np Michael. Thanks for dropping by. [12:00] Ubit Umarov: have fun michael.Christopher [12:00] Motoko.Karu @login.digiworldz.com:8002: bye [12:00] Michael.Christopher @cc.opensimulator.org:8002: see you all next week! [12:00] Andrew Hellershanks: We'll be here. :) [12:00] Arielle Popstar: No reason it cant be revisited Vincent [12:00] Arielle Popstar: things change [12:00] Vincent.Sylvester @hg.zetaworlds.com:8002: Bigger fish to fry [12:01] Andrew Hellershanks: Yes. [12:01] Arielle Popstar: always but they need more fire to get them heated before they fry [12:02] Ubit Umarov: and what news do you bring about opensim? :) [12:02] Arielle Popstar: Ubit could probably do the code in 10 minutes if he wanted to [12:02] Ubit Umarov: a few grids made it [12:02] Ubit Umarov: i mostly would need to just ask :p [12:02] Arielle Popstar: so 2 minutes to put in the premade patch :) [12:02] Andrew Hellershanks: Ubit, I think Vincent already covered most of the recent changes at the start of the hour. [12:03] Ubit Umarov: those where the ones i tried to detail [12:03] Arielle Popstar: any updates on Bullet MisterBlue? [12:03] Misterblue Waves: I'm releasing a new version of Convoar today that makes much more compliant GLTF files [12:03] Arielle Popstar: oh nice [12:04] Misterblue Waves: nothing done and nothing planned for BulletSim [12:04] Kayaker Magic: The bug I reported where a region hangs on some llCastRay calls, I'm upgrading one of the systems right now to see if Ubuntu 16.04 is the only place it fails. [12:04] Ubit Umarov: that is not a hang [12:04] Andrew Hellershanks: 16.04? That is a rather old version of Ubuntu. [12:04] Ubit Umarov: is taking some time [12:05] Ubit Umarov: raycast can be very slow [12:05] Ubit Umarov: on dense regions etc [12:05] Kayaker Magic: I noticed that it only fails near the SW corner in vars, so if some people can re-try again down there they may see the failure. [12:05] Ubit Umarov: fails? [12:05] Andrew Hellershanks: Any error messages? [12:06] Kayaker Magic: I call llCastRay on these systems and all avatars freeze and cannot move, all scripts in the region stop running for a minute. [12:06] Kayaker Magic: Is that a crash or a hang? [12:06] Ubit Umarov: you told nothing abotu fails.. just taking a lot of time [12:06] Ubit Umarov: yes that is not a fail [12:06] Ubit Umarov: it is take time [12:06] Misterblue Waves: for BulletSim, raycast is done all in software [12:07] Kayaker Magic: ubode and YEngine, as reported in the mantis, which I cannot add notes to at the moment.... [12:07] Ubit Umarov: a minute is a lot of time.. something odd [12:07] Ubit Umarov: but it is a current issue [12:08] Ubit Umarov: it takes time and holds a script thread for a lot of time [12:08] Kayaker Magic: I'm doing a bunch of new tests. Updating the operating system, planning a binary search of versions to find when it started. [12:08] Ubit Umarov: with X that meant X using more threads [12:08] Ubit Umarov: not on Y [12:08] Kayaker Magic: (versions of OpenSim) [12:09] Ubit Umarov: it looks worse with ubode ( runs on heartbeat) and Y engine.. ( has 2 exec threads) [12:09] Kayaker Magic: The discovery that it doesn't fail in the NE corner had me check for inside/ouside 256meters (var region issue) but I have seen failures outside that range. [12:10] Ubit Umarov: NE means nothin [12:10] Ubit Umarov: a ray as a start and a end [12:10] Ubit Umarov: NE is what.. start, end? [12:10] Kayaker Magic: Just this morning I think I noticed the time the avatars stay locked up seems to get less as I do the test farther from 0,0 [12:11] Kayaker Magic: As I move the prim around that runs the test script, in some places it locks up avatars, other locations it does not. [12:12] Kayaker Magic: I have a lot of tests to do, it will not get resolved by talking about it here..... [12:12] Ubit Umarov: rays and several other things need a new special code [12:12] Ubit Umarov: like ossl notecards [12:13] Kayaker Magic: I disable osMakeNotecard on all my regions. [12:13] Ubit Umarov: code that puts the script to sleep, and queues the work to be done by a external thread [12:13] Ubit Umarov: and taken out of sleep when result arrives [12:14] Kayaker Magic: (CreateNotecard? WriteNotecard? can't look up the name without opensimulator.org) [12:14] Ubit Umarov: in case of os notecards is actualy making them usind Dataserver [12:14] Ubit Umarov: as they should always had [12:14] Ubit Umarov: osnotceard where just bad ideas [12:15] Kayaker Magic: As God and Philip Linden intended! [12:15] Ubit Umarov: but i told that already :) [12:15] Arielle Popstar: probably a reason they did it that way [12:15] Ubit Umarov: dataserv is a external service thread [12:16] Ubit Umarov: that does the work async, not stoppinf the script threads [12:16] Ubit Umarov: osnotecards did by pass that, so script threads may stop waiting for web [12:17] Ubit Umarov: adding them was jsut ignorind the multhreading role of dataserv mechanics [12:17] Ubit Umarov: wle details.. [12:17] Ubit Umarov: and Yengine is in fact more sensitive to that [12:18] Ubit Umarov: so as i said, another mechanism is needed [12:18] Ubit Umarov: to enqueue work outside, putting that thread in sleep while it is donw [12:18] Ubit Umarov: ray cast also needs that [12:19] Ubit Umarov: bc those can only be done by physcis thread in a particular time slot [12:19] Ubit Umarov: currently once per heartbeat [12:20] Ubit Umarov: rays are a part of collision detection code [12:20] Ubit Umarov: or done when they are [12:20] Ubit Umarov: ( and yes.. can be very SLOW ) [12:21] Ubit Umarov: abuse of raycast == slow down region :) [12:21] Ubit Umarov: ( avatar cameras also do a lot ) [12:22] Ubit Umarov: wel when standing [12:22] Ubit Umarov: er and camera attached to the avatar [12:23] Selby.Evans @grid.kitely.com:8002: bye all [12:23] Andrew Hellershanks: Bye, Selby. [12:23] Ubit Umarov: cya selby.Evans [12:23] Motoko.Karu @login.digiworldz.com:8002: bye [12:23] Andrew Hellershanks: See you next week. [12:24] Kayaker Magic: I have to walk over to the barn to see how my update to Ubuntu 18.04 is going. [12:24] Kayaker Magic: see you all next week [12:24] Andrew Hellershanks: You keep it hidden in a barn? ;) [12:24] Motoko.Karu @login.digiworldz.com:8002: c cya [12:24] Misterblue Waves: take care all... bye bye [12:24] Andrew Hellershanks: ok, Kayaker. See you next week. [12:25] Andrew Hellershanks: It is almost half past the hour so it is about time to wrap up todays meeting unless there is any urgent last minute item(s) for today. [12:25] Arielle Popstar: Is Opensim in maintenance mode now? No new features considered or planned? [12:26] Andrew Hellershanks: The short answer to that would be no. [12:27] Arielle Popstar: so dying like s/l [12:27] Motoko.Karu @login.digiworldz.com:8002 whispers: ^^ [12:27] Vincent.Sylvester @hg.zetaworlds.com:8002: Then why are you here then? [12:27] Arielle Popstar: why are you? [12:27] Arielle Popstar: you bring nothing new? [12:27] Vincent.Sylvester @hg.zetaworlds.com:8002: Check the commit log and mantis once that is back up if you want to know what went on the last few months [12:28] Arielle Popstar: i have [12:28] Arielle Popstar: hence my question [12:28] Andrew Hellershanks: Before this goes downhill I'm going to call this meeting to a close. [12:28] Andrew Hellershanks: Thank you all for coming. See you again next week.