Réunion du 02-08-2022

Source : http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2022-08-02

Introduction

[11:00] Andrew Hellershanks : Bonjour, tout le monde.

[11:01] Andrew Hellershanks : Bonjour, Kayaker.

[11:01] Andrew Hellershanks : Bonjour, Selby.

[11:01] Kayaker Magic : Cette citation représente environ 15% de notre cerveau, basé sur 0% de science.

[11:02] Selby.Evans @grid.kitely.com:8002 : salut tout le monde.

[11:02] Kayaker Magic : Bonjour à tous !

[11:02] Ubit Umarov : Bonjour.

[11:02] Ubit Umarov : 15% c'est beaucoup ?

[11:02] Vincent.Sylvester @hg.zetaworlds.com:8002 : Je ne pense pas qu'il y ait eu de changements cette semaine, à part des choses faites sur les bibliothèques utilisées par OpenSim. Continuer à suivre les problèmes pour déterminer les causes et la canicule qui rend tout le monde somnolent.

[11:02] Kayaker Magic : Eh bien, peut-être basé sur l'observation de beaucoup de gens....

[11:03] Ubit Umarov : oui, j'ai joué un peu avec le code natif ubode.

[11:03] Ubit Umarov : j'ai finalement réussi à faire fonctionner le code compilé avec un ancien gcc.

[11:03] Ubit Umarov : il a commencé à s'embrouiller avec certaines choses du code légal...........

Mono

[11:04] Vincent.Sylvester @hg.zetaworlds.com:8002 : J'ai eu une discussion amusante avec les gars responsables de Mono sur leur discord, confirmant essentiellement ce que nous savions déjà qu'ils ne se soucient pas vraiment de continuer le travail sur Mono. Heureusement, étant donné qu'il s'agit d'un logiciel libre, la communauté qui s'en préoccupe peut continuer le travail, donc pas de raison de s'inquiéter.

[11:04] Vincent.Sylvester @hg.zetaworlds.com:8002 : Microsoft : Acheter, vider, tuer comme d'habitude

[11:05] Ubit Umarov : on s'y attendait un peu, ce sont eux qui poussent pour .net7 aussi.

[11:05] Ubit Umarov: :)

[11:05] Ubit Umarov : aucune idée de ce qui est arrivé aux anciens responsables mono.

[11:05] Vincent.Sylvester @hg.zetaworlds.com:8002 : J'avais très peu de respect pour Microsoft et ses employés, mais ils ont atteint de nouveaux sommets.

[11:05] Ubit Umarov : je suppose qu'ils ont reçu de bons revenus et qu'ils n'ont pas pris sa retraite ?

[11:05] Ubit Umarov : aucune idée.

[11:06] Kayaker Magic : J'ai entendu dire qu'ils ont tous été engagés par MicroSquishy.

[11:06] Ubit Umarov : fautes de frappe.

[11:06] Vincent.Sylvester @hg.zetaworlds.com:8002 : Probablement, juste après avoir mis MS sur le CV pour obtenir ce joli salaire de consultant.

[11:06] Ubit Umarov : "J'imagine qu'ils ont reçu beaucoup d'argent et qu'ils sse sont retirés ?"

[11:08] Ubit Umarov : mais bon, mono fonctionne toujours.

[11:08] Vincent.Sylvester @hg.zetaworlds.com:8002 : Le repo apt qu'ils maintiennent est un peu plus désordonné que je ne le pensais car il ne se nettoie pas , donc tout le code déprécié est laissé sur les machines nécessitant soit une réinstallation soit un nettoyage manuel des vieux trucs, parce que si vous ne le faites pas, il fait de son mieux pour les recharger.

Fork NGC d'OpenSim

[11:09] Kayaker Magic : Ce fork NGC d'OpenSim, ils ont réussi à le compiler et à le faire fonctionner avec . net sur des serveurs MicroSquishy, mais pas avec . net sur des serveurs linux.

[11:09] Ubit Umarov : non, ce n'est pas cela.

[11:10] Ubit Umarov : c'est .net4.8

[11:10] Vincent.Sylvester @hg.zetaworlds.com:8002 : L'obtention de la compilation et son exécution correcte sont des mondes à part. J'ai réussi à le compiler sur ReactOS il y a quelque temps, mais il a explosé au moment de l'exécution.

[11:10] Ubit Umarov : aucune idée de ce qu'ils ont changé par rapport à notre code, à part le nom.

[11:11] Ubit Umarov : et ils ont plusieurs projets.

[11:11] Ubit Umarov : l'un d'eux est vraiment de passer à Dotnet.

[11:12] Vincent.Sylvester @hg.zetaworlds.com:8002 : Maintenir un fork sur de simples correctifs ou modules ne mène généralement à rien, la charge de travail est beaucoup plus élevée que ce que la plupart des gens anticipent.

[11:13] Andrew Hellershanks : Bonjour, Michael

[11:13] Vincent.Sylvester @hg.zetaworlds.com:8002 : La fusion rétroactive avec le noyau demande beaucoup trop d'efforts je trouve.

[11:13] MichaelOmega ChristopherGeode : bonjour.

[11:13] Vincent.Sylvester @hg.zetaworlds.com:8002 : Ajouter des choses par dessus est beaucoup plus facile

[11:13] MichaelOmega ChristopherGeode : j'ai abandonné et créé un avatar sur OSgrid pour arriver ici, je ne sais toujours pas comment le faire depuis ossc grid :)

[11:14] Vincent.Sylvester @hg.zetaworlds.com:8002 : De plus, il y a les structures de modules pour tout ce qu'on veut ajouter et qui ne nécessite pas de changements profonds.

[11:14] Vincent.Sylvester @hg.zetaworlds.com:8002 : Ces modules peuvent même écraser la plupart des modules intégrés, mais pas tous.

[11:14] MichaelOmega ChristopherGeode : et salut vincent et andrew, ça fait trop longtemps lol

[11:14] Andrew Hellershanks : Michael, content que tu sois là.

[11:15] MichaelOmega ChristopherGeode : merci.

[11:16] Vincent.Sylvester @hg.zetaworlds.com:8002 : Le développement, quel qu'il soit, est une bonne chose, mieux que la stagnation, il crée au moins un dialogue... en quelque sorte...

[11:18] Vincent.Sylvester @hg.zetaworlds.com:8002: Having done from scratch development on Mono though I don't see what supposedly is bad about it. It's actually quite easy to get an application off the ground and working cross platform. You can compile with msbuild and it will run on Mono or .net or you just mkbundle Mono into it and it becomes fully portable without any dependencies. [11:20] Andrew Hellershanks: Good to know there are people out there that still care about Mono and are willing to continue working on it. [11:21] Vincent.Sylvester @hg.zetaworlds.com:8002: Things done with it work, with .net to Mono you never really know, but Mono to .net tends to just work, which makes things a joy to write [11:21] Vincent.Sylvester @hg.zetaworlds.com:8002: Less headache over whether stuff is going to blow up at runtime, even things like filesystem are surprisingly easy to deal with [11:21] Ubit Umarov: well only a limited set of .net "universe" does run on mono [11:22] Ubit Umarov: and a lt was only fixed when ms did get in, in version 6 [11:22] MichaelOmega ChristopherGeode: Hi Ubit, somehow didnt see you lol [11:22] Ubit Umarov: ofc ms did that so to use it as core of dotnet for linux [11:23] Ubit Umarov: names like monoCLR there [11:23] Ubit Umarov: named [11:23] Vincent.Sylvester @hg.zetaworlds.com:8002: Buying Mono, gutting it for the good stuff, then just leaving it to die in the dust, MS being MS [11:23] Ubit Umarov: well they also stopped .net4.8 [11:24] Vincent.Sylvester @hg.zetaworlds.com:8002: The Electronic Arts of enterprise software [11:24] Misterblue Waves: I think the main things keeping OS from converting to dotnet are the dependent libraries (libomv, ...) [11:24] Ubit Umarov: and that is a lot of opensim :p [11:24] Ubit Umarov: and lost xengine, etc [11:24] Andrew Hellershanks nods [11:25] Andrew Hellershanks: Just like we have occasionally (jokingly) talked about rewriting OS in something other than C#. [11:25] Ubit Umarov: not clear if Yengine will work also [11:26] Misterblue Waves: it's a gigantic job, but then, if we'd started it 10 years ago when we first talked about it, it would be more than done now :) [11:26] Ubit Umarov: everything that is outside of what ms team considers a normal app... whatever that is, is in troubler [11:26] Misterblue Waves: what might yengine be using that isn't in dotnet [11:26] Misterblue Waves: ? [11:27] Ubit Umarov: dynamic code [11:27] Ubit Umarov: is not "normal application use" [11:28] Ubit Umarov: on a blog one of those did check the use of image code on normal apps, and found it was not used, except on a few thigns to gen QR codes.." [11:28] Vincent.Sylvester @hg.zetaworlds.com:8002: You'd have to effectively create some sort of virtualization to "run" things in and at that point you are back to a runtime, which is what Mono is so re-inventing the wheel [11:29] Ubit Umarov: so.. image is not part of normal applications dor them, so not suported [11:29] Ubit Umarov: :p [11:29] Vincent.Sylvester @hg.zetaworlds.com:8002: Overgeneralization of the century there lol [11:29] Ubit Umarov: well t be honest the major issue they have with it is libgdi++ [11:29] Ubit Umarov: the mono hack to replace real windows gdi [11:30] Vincent.Sylvester @hg.zetaworlds.com:8002: C# is quite commonly used for dynamic code actually. I recall writing some C# plugins for some games that dynamically loaded as soon as you saved your code, which makes development really easy [11:30] Ubit Umarov: that does depend on a lot of other linux lins like pango [11:30] Vincent.Sylvester @hg.zetaworlds.com:8002: Some game engines are built with that in them as well [11:30] Ubit Umarov: in fact there is at least a incompatibility on how alpha is handled by pango [11:31] Ubit Umarov: i had a small fight with that on dyn textures [11:31] Ubit Umarov: and as you remember linux was doing alpha the opositite of windows [11:31] Ubit Umarov: wlel od something like that :) [11:32] Ubit Umarov: so yeah libgdi++ always had issues [11:32] Vincent.Sylvester @hg.zetaworlds.com:8002: If that code was easy to modify or change we'd have dynamic textures on existing textures already, blending them together, but oh boy that code is spaghetti [11:32] Ubit Umarov: so they jsut refuse to try to mantain it [11:32] Vincent.Sylvester @hg.zetaworlds.com:8002: One of those things you gotta write in one go while high on energy drinks and coffee [11:33] Ubit Umarov: and so no Imagor code for linux [11:33] Ubit Umarov: meanwhile they added it back, windows only [11:33] Ubit Umarov: so 99% of computer are happy :p [11:33] MichaelOmega ChristopherGeode: rofl@energy drinks and coffee - the coffee part is too weak, it just slows you back down ;) [11:35] Vincent.Sylvester @hg.zetaworlds.com:8002: Dynamic textures ideally like bakes should be done on the viewer, just receiving the calls. Viewer usually has GPU... you know.. a graphics processing unit... but that's asking the clouds from sky [11:35] Ubit Umarov: i also found a few diferences on dotnet APIs [11:35] Misterblue Waves: I'd still like to find the time to make a libomv-lite that could be made dotnet. There is a lot in that library OS doesn't use [11:35] Ubit Umarov: suposed to be same as .net... but not really [11:35] Ubit Umarov: remember .net core 1 was made clear as a big breaking change [11:36] Ubit Umarov: as a brave new thing [11:36] Misterblue Waves: then to start re-working OS to just dotnet-6 it [11:36] Ubit Umarov: guess it took ms management to step in and force them [11:36] Ubit Umarov: so in Core 3 then made a major efford to recover some compatibilty [11:37] Misterblue Waves noticing that "dotnet-6" is too close to "deep 6"ing it ;-) [11:37] Ubit Umarov: such was that they where forced to rename it .Net5 [11:37] Andrew Hellershanks grins at Misterblue [11:37] Ubit Umarov: telling it was just a name change lol [11:37] Vincent.Sylvester @hg.zetaworlds.com:8002: Never mind that I am not convinced that even MS knows where they want to go... where is the oh so great .NET Enterprise that supposedly was a thing they were building... yeah gotta wait for them to make up their mind of crash the cart into the wall first before committing something as large as OpenSim to a course [11:37] Ubit Umarov: nahh it was to try to clear the image that Core was a major breaking change [11:38] Ubit Umarov: if you read Core early docs, it as just a brave new thing.. now a upgrade to .net [11:39] Ubit Umarov: but alot stayed out [11:39] Ubit Umarov: and a lot looks the same, but it is not.. [11:39] Ubit Umarov: bahh [11:39] Ubit Umarov: ofc not impossible to convert [11:40] Ubit Umarov: like several more months of unpaid work for no good reason... [11:40] Ubit Umarov: with hard debug [11:40] Ubit Umarov: see? [11:40] Vincent.Sylvester @hg.zetaworlds.com:8002: I wonder who thought about that in the first place. Think it's fairly clear these days that most rather keep legacy stuff around than spend the time to re-develop unless there is a clear benefit that can be quantified into cash [11:40] Ubit Umarov: and it can't be a partial change [11:40] Vincent.Sylvester @hg.zetaworlds.com:8002: Banks still running on systems from the 70's [11:41] Ubit Umarov: it will take a lot of code fix, bc anything starts tiking again [11:41] Vincent.Sylvester @hg.zetaworlds.com:8002: Even SAP still has tons of code written before I was born [11:41] Ubit Umarov: bc == before [11:41] Ubit Umarov: olf code is good [11:41] Ubit Umarov: CERN stayed in fortran bc of that [11:41] Andrew Hellershanks: :) [11:42] Ubit Umarov: they had tons and tons of just the best math code made [11:42] Vincent.Sylvester @hg.zetaworlds.com:8002: Re-inventing the wheel just for the sake of say being able to spell wheel with one less e isn't really a good use of time and resources [11:42] Ubit Umarov: losing all that just bc useless languages changes?? [11:42] Ubit Umarov: ppl real good programmers are a rare thing.. [11:43] Ubit Umarov: most code is pure trash..that kinda does things.. [11:43] Ubit Umarov: you do not send to trash code made by real good coders.. [11:43] Vincent.Sylvester @hg.zetaworlds.com:8002: Quantifiable difference is key, just upgrading to newer standards when it ends up doing the same things in the same time is wasted energy [11:43] Ubit Umarov: specialy on not trivial math code with error handling etc [11:44] Kayaker Magic: I have a saying: "A bad programmer can write FORTRAN in any language" [11:44] Andrew Hellershanks: hehe [11:44] Ubit Umarov: btw fortran bc it was really the best to produce optimized code [11:45] Misterblue Waves: the part of OS that scares me is all that modular plumbing mel and diva put in for all the versions of hg/robust/etc... that is a tangle of interlinked, dynamic loading dlls with just enough exceptions for "performance" [11:45] Ubit Umarov: its rigid structure allowed for compilers to make less "mistakes" [11:45] Ubit Umarov: yeah those modules.. well [11:46] Vincent.Sylvester @hg.zetaworlds.com:8002: I was looking at monitoring stuff, something to show load average and memory, kept getting results for all manners of projects most using graphana which is a pain in the neck to get working right natively. rrdtool and collectd are ancient by todays standards, but work just fine to show the data and are lightweight. [11:46] Ubit Umarov: we still have that need for postPostPostInit? [11:46] Vincent.Sylvester @hg.zetaworlds.com:8002: Look at Gitlab, started as replacement for bloat and slow bitbucket and github and now itself is borderline bloatware that is becoming slow and annoying to deal with [11:46] Misterblue Waves: OS has more than one stats systems LOL [11:47] Ubit Umarov: well remmeber os was made as a framework [11:47] Vincent.Sylvester @hg.zetaworlds.com:8002: Did extend jsonsimstats a bit actually so the others are almost pointless now [11:47] Ubit Umarov: in fact should not even run [11:47] Ubit Umarov: we have that stated on some docs.. [11:48] Ubit Umarov: only a few crazy devs decided opensim should try to do something :p [11:48] Ubit Umarov: ( wel the author of those docs also did ) [11:49] Vincent.Sylvester @hg.zetaworlds.com:8002: Still a lot easier to fiddle with than some other code doing far less [11:49] Ubit Umarov: on modules the most irritating situation for me now is grid login [11:49] Ubit Umarov: there is none [11:49] Andrew Hellershanks: Vincent, I like to be able to see stats without having to mentally parse json code. [11:50] Vincent.Sylvester @hg.zetaworlds.com:8002: Well the data from json things is not really useful for anything statistics related as it isn't reliable still [11:50] Vincent.Sylvester @hg.zetaworlds.com:8002: Scripts count keeps jumping to weird numbers [11:50] Ubit Umarov: a region is only checked if allowed when it is all loaded even map uploaded [11:50] Vincent.Sylvester @hg.zetaworlds.com:8002: Still haven't found what's fubar with that [11:50] Ubit Umarov: and we can't do it that earlier [11:51] Ubit Umarov: because the region does not have any comms modules loaded until that stage mosttly [11:51] Ubit Umarov: bc cooms modules are like all the others [11:51] Andrew Hellershanks: why is jsonsimstats not gettings its data from the sample place(s) as used by webstats? [11:52] Vincent.Sylvester @hg.zetaworlds.com:8002: I think it mostly does, script count is internally broken somewhere, cause sim status in lsl get the same wrong numbers [11:52] Ubit Umarov: like MB told we have several stats engines [11:52] Vincent.Sylvester @hg.zetaworlds.com:8002: I haven't started digging too deep cause it's not a critical thing, just a stat [11:53] Ubit Umarov: come on.. opensim has no fingers,.. how can it count scripts? [11:53] Vincent.Sylvester @hg.zetaworlds.com:8002: Avatar count also reads NPCs still and I am not quite sure why [11:53] Vincent.Sylvester @hg.zetaworlds.com:8002: I recall the fix for that, but might not have been enough or missing a flag [11:53] Vincent.Sylvester @hg.zetaworlds.com:8002: If the day had 48 hours I might get around to it [11:54] Andrew Hellershanks: That makes the point of NPC count should be available as a separate stat. Sounds like a flag isn't being checked somewhere in avatar counting. [11:54] MichaelOmega ChristopherGeode: ^ that right there. there just arent enough hours in the day [11:54] Misterblue Waves: jsonsimstats was built so one could have a scripted web page to display the stats (JS that reads the JSON and builds tables for display) [11:54] Ubit Umarov: thnkg is that some devs did consider we should see no dif btw npcs and avatars [11:55] Vincent.Sylvester @hg.zetaworlds.com:8002: Andrew that actually is in there now, but something is still not quite working as intended [11:55] Ubit Umarov: so not we need go trace all spots where there should be a diference [11:55] Vincent.Sylvester @hg.zetaworlds.com:8002: Did change this what almost two years ago now I think [11:55] Ubit Umarov: remember NPCs where added for same reason as bots in SL [11:55] Ubit Umarov: to pretend that regions had people [11:55] Vincent.Sylvester @hg.zetaworlds.com:8002: It checks for obvious NPCs, but I suspect if they are marked to be treated as avatars it still counts them, something like that [11:56] Ubit Umarov: well that are counted diferently in most places now [11:56] Ubit Umarov: ( bc a dev arrived that hates NPCs as human replacements :P ) [11:57] Andrew Hellershanks nods [11:57] Misterblue Waves: on a different subject... does anyone remember someone building an S3 based asset storage system? [11:58] Ubit Umarov: s3 ? [11:58] Vincent.Sylvester @hg.zetaworlds.com:8002: To do what? Introduce more layers of failure to it? [11:58] MichaelOmega ChristopherGeode: i know someone talked about an S3 system at one point - dont know if it got built by anyone else [11:58] Andrew Hellershanks: The stats code shouldn't have to work out how to count avatars/NPCs as here should be a core callable function that would return the correction information (possibly requiring a flag to say whether to include NPCs or not). [11:58] Selby.Evans @grid.kitely.com:8002: Must go -- Bte all [11:58] MichaelOmega ChristopherGeode: bye selby! [11:58] Ubit Umarov: cya [11:59] Misterblue Waves: bye selby [11:59] Andrew Hellershanks: Bye, Selby [11:59] Ubit Umarov: amazon s3 ? [11:59] MichaelOmega ChristopherGeode: ubit - yea [11:59] Vincent.Sylvester @hg.zetaworlds.com:8002: Andrew that's the easier to fix part I imagine, given we do track that properly, but I think the stat code just grabs the wrong data or something like that. Bigger problem is active scripts which can be all over the place, even negative for some reason [11:59] Misterblue Waves: some s3 interfaced DB -- there are a lot of S3 storage servioces that are WAY cheaper than Amazon [12:00] Ubit Umarov: like what sl uses? [12:00] Ubit Umarov: mb kitely is using something like that also [12:00] Ubit Umarov: fun SL now is as secure as MOSES told opensim was not [12:00] Ubit Umarov: LOL [12:01] MichaelOmega ChristopherGeode: lol [12:01] Ubit Umarov: moving to systems like that turns all assets public and unprotected [12:01] Ubit Umarov: at least real implementations in use [12:02] Ubit Umarov: ofc in our case also kills local assets like dyn textures [12:02] MichaelOmega ChristopherGeode: because encryption takes too much time/reseources to use something actually secure? [12:02] Ubit Umarov: no.. anyone can get them even without any login [12:02] Jamie.Jordan @grid.kitely.com:8002: Have a great week yall [12:03] MichaelOmega ChristopherGeode: you too jamie [12:03] Andrew Hellershanks: You too, Jamie. [12:03] Andrew Hellershanks: We are a little past the top of the hour. Any final questions/comments for today? [12:03] Ubit Umarov: see the old MOSES thing about osgrid security.. it now means SL also :P [12:04] Ubit Umarov: they did kept a few things on LLUDP so kinda still safe [12:04] Vincent.Sylvester @hg.zetaworlds.com:8002: Security by obscurity that well not entirely [12:05] Ubit Umarov: that is the only one that actually works [12:05] Ubit Umarov: LOL [12:05] Ubit Umarov: but nothing about obscurity [12:06] Ubit Umarov: is access control.. or lack of any [12:07] Andrew Hellershanks: If there are no other questions/comments for today I will call this gathering to a close. My cat will be happy about that. She is looking for attention from me. :) [12:07] Andrew Hellershanks: Thank you all for coming. See you again next week. [12:07] Kayaker Magic: Buy all! [12:07] MichaelOmega ChristopherGeode: I think I'm good - lots of thoughts/projects but nothing ready to share yet [12:07] MichaelOmega ChristopherGeode: bye everyone!