|
|
| (29 versions intermédiaires par le même utilisateur non affichées) |
| Ligne 5 : |
Ligne 5 : |
| {{Avertissement_résumé|fond=pink |bord=red |message = Ce résumé existe pour orienter vos recherches. Des erreurs d'interprétation ne sont pas à exclure. Pour plus de précisions, veuillez vous référer aux sources ou vous adresser directement aux développeurs d'OpenSimulator en assistant aux [http://opensimulator.org/wiki/Office_hours réunions du mardi] ou sur [http://opensimulator.org/wiki/IRC le canal IRC]}} | | {{Avertissement_résumé|fond=pink |bord=red |message = Ce résumé existe pour orienter vos recherches. Des erreurs d'interprétation ne sont pas à exclure. Pour plus de précisions, veuillez vous référer aux sources ou vous adresser directement aux développeurs d'OpenSimulator en assistant aux [http://opensimulator.org/wiki/Office_hours réunions du mardi] ou sur [http://opensimulator.org/wiki/IRC le canal IRC]}} |
|
| |
|
| = Noyau= | | = Code général = |
| = Base de données = | | === Utilisation du CPU === |
| = Modules = | | * L'utilisation de Dotnet provoquerait une utilisation plus importante du CPU, mais tout le monde n'a pas vu d'augmentation spéctaculaire. |
| | * Observation de Vincent Sylvester : Augmentation probable de l'ordre de 15 à 30 % selon les cas. Il a remarqué que c'était un peu linéaire jusqu'à un certain point et que l'utilisation a tendance à augmenter beaucoup plus rapidement puis à ralentir un peu. "C'est effrayant de voir une utilisation de 250% cpu avec beaucoup d'utilisateurs." |
| | * Problème potentiel : Quelque chose qui tourne deux fois plus vite sans aucune limitation pourrait causer des goulots d'étranglement ailleurs. |
| | * NDLR 🏗️ (je sature il reste encore 6 minutes de chat.) |
| | |
| = Bugs = | | = Bugs = |
| === Les librairies non gérées de macOS ne se chargent pas === | | === Les librairies non gérées de macOS ne se chargent pas === |
| Ligne 16 : |
Ligne 20 : |
| * Précision de Gavin Hird : si un logiciel est démarré depuis le terminal, la signature n'est pas vérifiées. | | * Précision de Gavin Hird : si un logiciel est démarré depuis le terminal, la signature n'est pas vérifiées. |
| * NDLR : je n'ai jamais utilisé un Mac de ma vie, donc je vais m'arrêter là dans le résumé de ce sujet de peur de faire de grosses erreurs. | | * NDLR : je n'ai jamais utilisé un Mac de ma vie, donc je vais m'arrêter là dans le résumé de ce sujet de peur de faire de grosses erreurs. |
| | ===Disparition de la bibliothèque OpenSim === |
| | * La bibliothèque OpenSim aurait disparu dans l'inventaire du viewer (NDLR: sur Argentoratum c'est le cas). |
| | * Ubit Umarov a encore la bibliothèque même dans Firestorm Beta avec les récents changements. |
| | * Le problème doit venir de la configuration Robust. (NDRL : je devais regarder ce qui se passe, je mettrai la solution ici si je la trouve). |
|
| |
|
| = Tests =
| |
| = Projets en cours / Infos=
| |
| = Viewers= | | = Viewers= |
| === Dayturn === | | === Dayturn === |
| * Compilation du viewer Windows avec Visual Studio 2022 | | * Compilation du viewer Windows avec Visual Studio 2022 |
| | * Gavin Hird a pour projet d'essayer Mumble avec la version Mac du viewer. |
|
| |
|
| === Firestorm === | | === Firestorm === |
| * Avec la version 7.0 alpha certaines choses ne fonctionnent plus à cause d'un changement de protocole d'inventaire. Des vignettes d'inventaire s'affichnet mais ne fonctionnent pas parce que les développeurs utilisent le code pour [[Réunion_du_03-10-2023#Viewer_et_AIS |l'AIS V3 (voir la réunion du 03-10-2023)]]. | | * Avec la version 7.0 alpha certaines choses ne fonctionnent plus à cause d'un changement de protocole d'inventaire. Des vignettes d'inventaire s'affichent mais ne fonctionnent pas parce que les développeurs utilisent le code pour [[Réunion_du_03-10-2023#Viewer_et_AIS |l'AIS V3 (voir la réunion du 03-10-2023)]]. |
|
| |
|
| === Second Life === | | === Viewer Second Life === |
| * Il semble que SL travaille sur beaucoup de choses à la fois et tout devient instable dans les viewers. | | * Il semble que SL travaille sur beaucoup de choses à la fois et tout devient instable dans les viewers. |
| * Gavin.Hird est presque sûr que SL developpe une version Xbox du viewer. Ils pourraient aussi abandonner le support Mac, parce qu'ils ont mis à jour les instructions de construction pour la version Windows, mais les instructions de la version Mac sont significativement dépassées. Les utilisateurs de Mac seront à la traîne par rapport aux utilisateurs de Windows en termes de fonctionnalités et de performances dixit le chef technique de SL. Apple a dit qu'ils allaient se débarrasser d'OpenGL et la seule chose qui fonctionne sur macOS est Metal. | | * Gavin.Hird est presque sûr que SL developpe une version Xbox du viewer. Ils pourraient aussi abandonner le support Mac, parce qu'ils ont mis à jour les instructions de construction pour la version Windows, mais les instructions de la version Mac sont significativement dépassées. Les utilisateurs de Mac seront à la traîne par rapport aux utilisateurs de Windows en termes de fonctionnalités et de performances dixit le chef technique de SL. Apple a dit qu'ils allaient se débarrasser d'OpenGL et la seule chose qui fonctionne sur macOS est Metal. |
| Ligne 33 : |
Ligne 40 : |
| * Paradoxalement, il est aussi question d'un viewer pour iPad. | | * Paradoxalement, il est aussi question d'un viewer pour iPad. |
| * Le projet du Viewer mobile fonctionne avec [https://fr.wikipedia.org/wiki/Vulkan_(API) Vulkan]. Cuga.Raja suppose qu'UDP ne sera plus utilisé (donc passage à TCP) pour le Viewer mobile. L'ancienne idée pour le mobile était d'utiliser une sorte de streaming vidéo côté serveur. | | * Le projet du Viewer mobile fonctionne avec [https://fr.wikipedia.org/wiki/Vulkan_(API) Vulkan]. Cuga.Raja suppose qu'UDP ne sera plus utilisé (donc passage à TCP) pour le Viewer mobile. L'ancienne idée pour le mobile était d'utiliser une sorte de streaming vidéo côté serveur. |
| * Les viewers OpenSim sont à 95% du code Linden Lab. D'après Ubit Umarov, la corde de la dépendance est peut-être sur le point de se rompre. Certains changements sont tout simplement trop mauvais pour être suivis. Quand firestorm a speraté OpenSim et SL, ils ont arrêté de faire du code pour nous... | | * Les viewers OpenSim sont à 95% du code Linden Lab. D'après Ubit Umarov, la corde de la dépendance est peut-être sur le point de se rompre. Certains changements sont tout simplement trop mauvais pour être suivis. Quand firestorm a speraté OpenSim et SL, ils ont arrêté de faire du code pour nous... Vincent Sylvester dit changer le pipeline de développement et nettoyer le désordre des dépendances ne serait pas seulement bénéfique pour nous, mais rendrait généralement plus facile le développement d'un viewer en utilisant le code Linden Lab. C'est une entreprise de grande envergure et cela ne résout pas complètement le problème de la compatibilité. |
| * Le problème est que personne n'est capable de réécrire le moteur de rendu. | | * Le problème est que personne n'est capable de réécrire le moteur de rendu. |
|
| |
|
| Ligne 39 : |
Ligne 46 : |
| * Sharpview est 0% de code Linden Lab et 100% [https://fr.wikipedia.org/wiki/Rust_(langage) Rust]. | | * Sharpview est 0% de code Linden Lab et 100% [https://fr.wikipedia.org/wiki/Rust_(langage) Rust]. |
| * Lorsque quelqu'un veut le code, il est dirigé vers un test open source de la pile de rendu, et il doit le compiler et de le construire. Jusqu'à présent, personne ne l'a fait. | | * Lorsque quelqu'un veut le code, il est dirigé vers un test open source de la pile de rendu, et il doit le compiler et de le construire. Jusqu'à présent, personne ne l'a fait. |
| * '''Pour tester ,ce lien mène vers un jeu factice avec la même structure que Sharpview''' et qui doit être compilé : https://github.com/John-Nagle/ui-mock | | * '''Test : ce lien mène vers un jeu factice avec la même structure que Sharpview''' et qui doit être compilé : https://github.com/John-Nagle/ui-mock . |
| | ** Tout est géré par le système [https://fr.wikipedia.org/wiki/Cargo_(informatique) Cargo] de Rust ([https://doc.rust-lang.org/stable/cargo/ Site officiel(en)]. Il suffit d'installer Rust, d'essayer "cargo build", et de voir ce qui se passe. |
| | ** Ce test fonctionne sous Windows et Linux. |
| | ** Il faut encore tester sous MacOS. Faites savoir à Joe Magarac comment cela fonctionne. Il devra être compilé dans Xcode pour que l'application finale s'exécute comme une application de bureau, donc idéalement il doit être signé. |
|
| |
|
| | = Compilation = |
| | * Il a été question de compilation [https://fr.wikipedia.org/wiki/Compilation_anticip%C3%A9e#Articles_connexes AOT (Ahead-of-Time)] et [https://fr.wikipedia.org/wiki/Compilation_%C3%A0_la_vol%C3%A9e JIT (Just-in-Time)]. Il semble qu'OpenSim ait choisi la compilation JiT . Le changement de performance entre AOT et JIT est principalement sur le chargement des méthodes. |
| = Source= | | = Source= |
| http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2023-11-07 | | http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2023-11-07 |
| <!-- | | <!-- |
| [12:05 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: any thoughts progress on voice?
| |
| [12:06 PM PT] Cuga.Rajal @rajal.org:9000: Well, and urgent but easy fix issue came up yesterday, with the macOS unmanaged libraries. I opened a Mantis and attached some updated libs
| |
| [12:07 PM PT] Gavin.Hird @grid.xmir.org:8002: The only thing I have done the last week is to get the Windows viewer to build with Visual Studio 2022
| |
| [12:07 PM PT] Orbert.Tatham @hg.zetaworlds.com: Quite an accomplishment, that
| |
| [12:07 PM PT] Gavin.Hird @grid.xmir.org:8002: plus some other build issues on macOS because Apple has been creative again
| |
| [12:07 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: dayturn?
| |
| [12:07 PM PT] Gavin.Hird @grid.xmir.org:8002: yes
| |
| [12:08 PM PT] Cuga.Rajal @rajal.org:9000: opensim is non-op on macOS until the new libs are swapped in to to dotnet6 branch
| |
| [12:08 PM PT] Gavin.Hird @grid.xmir.org:8002: is it?
| |
| [12:08 PM PT] Cuga.Rajal @rajal.org:9000: yep
| |
| [12:08 PM PT] Andrew Hellershanks: Cuga, The change log shows the libraries as having been pushed to the code repo.
| |
| [12:08 PM PT] Gavin.Hird @grid.xmir.org:8002: I have not had any issues
| |
| [12:09 PM PT] Cuga.Rajal @rajal.org:9000: great
| |
| [12:09 PM PT] Cuga.Rajal @rajal.org:9000: then macOS people will need to update
| |
| [12:09 PM PT] Gavin.Hird @grid.xmir.org:8002: if you start anything from terminal the signature isn't even checked
| |
| [12:09 PM PT] Jagga Meredith: I'm now a MacOS people. God fed up with Windoze.
| |
| [12:09 PM PT] Cuga.Rajal @rajal.org:9000: no ttrue
| |
| [12:09 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: this affects just nessie or all builds?
| |
| [12:10 PM PT] Cuga.Rajal @rajal.org:9000: macOS specifically, dotnet6 branch
| |
| [12:10 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: ty
| |
| [12:11 PM PT] Gavin.Hird @grid.xmir.org:8002: what version of macOS?
| |
| [12:11 PM PT] Cuga.Rajal @rajal.org:9000: it was a code signing issue, may have been my fault
| |
| [12:11 PM PT] Cuga.Rajal @rajal.org:9000: The libs will run on any macOS 10.15-114.x
| |
| [12:11 PM PT] Cuga.Rajal @rajal.org:9000: but code signing issue will break on all of them
| |
| [12:12 PM PT] Cuga.Rajal @rajal.org:9000: oops I mean 10.5-latest
| |
| [12:12 PM PT] Cuga.Rajal @rajal.org:9000: 10.15-latest
| |
| [12:12 PM PT] Gavin.Hird @grid.xmir.org:8002: all right, you say you cannot sign it any more
| |
| [12:12 PM PT] Cuga.Rajal @rajal.org:9000: I can sign it, and did
| |
| [12:12 PM PT] Gavin.Hird @grid.xmir.org:8002: but the old ones runs perfecty fine
| |
| [12:12 PM PT] Cuga.Rajal @rajal.org:9000: the problem has to do with the developer cerytificate needing to be renewed annually
| |
| [12:13 PM PT] Cuga.Rajal @rajal.org:9000: The certificate I used to sign the macOS libs was originally created November 2022, and I got an email from Apple that they were going to expire and I needed to create new ones
| |
| [12:13 PM PT] Gavin.Hird @grid.xmir.org:8002: sure but the signature is never checked if it is just a lib running in some other process under BSD
| |
| [12:14 PM PT] Cuga.Rajal @rajal.org:9000: as part of creating new ones, I had to revoke the old cert, whiuch is the one used to sign the macOS unmanaged libs
| |
| [12:14 PM PT] Gavin.Hird @grid.xmir.org:8002: when did you revoke it?
| |
| [12:15 PM PT] Cuga.Rajal @rajal.org:9000: The day after I revoked the old cert, my server auto-retarts oensim evenry Monday morning, and it failed to start. Could not load any of the 3 unmanaged libs
| |
| [12:15 PM PT] Cuga.Rajal @rajal.org:9000: same result on x86_64 Mac and arm64 Mac
| |
| [12:15 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: hi kayaker
| |
| [12:15 PM PT] Cuga.Rajal @rajal.org:9000: So I re-signed the same binaries with the new cert, dropped them in, and it worked again
| |
| [12:15 PM PT] Gavin.Hird @grid.xmir.org:8002: starting a simmulator now. not a pip
| |
| [12:15 PM PT] Gavin.Hird @grid.xmir.org:8002: with the old libs
| |
| [12:15 PM PT] Cuga.Rajal @rajal.org:9000: hmm, no idea why your result is different
| |
| [12:16 PM PT] Cuga.Rajal @rajal.org:9000: glad to hear though
| |
| [12:16 PM PT] Ubit Umarov: i did push the new libs for mac, a hour or so ago
| |
| [12:16 PM PT] Gavin.Hird @grid.xmir.org:8002: I have an idea, lemme check
| |
| [12:16 PM PT] Ubit Umarov: and those where last week "code changes"
| |
| [12:16 PM PT] Cuga.Rajal @rajal.org:9000: it did give me a scare, that others are having same problem, so I re-signed them and pushed the new ones to Mantis
| |
| [12:17 PM PT] Cuga.Rajal @rajal.org:9000: we need to understand this better. ideally should not have to re-sign unmanaged libs once a year
| |
| [12:17 PM PT] Cuga.Rajal @rajal.org:9000: Gavin, perhaps we can look into this tigether away from meeting
| |
| [12:17 PM PT] Gavin.Hird @grid.xmir.org:8002: sure
| |
| [12:18 PM PT] Cuga.Rajal @rajal.org:9000: so if anyone usins macOS is having trouble starting up opensim /dotnet6 just need to update
| |
| [12:18 PM PT] Jagga Meredith: k
| |
| [12:18 PM PT] Cuga.Rajal @rajal.org:9000: thanks for pushing those into repo
| |
| [12:19 PM PT] Ubit Umarov: well thx for providign them :)
| |
| [12:20 PM PT] Cuga.Rajal @rajal.org:9000: stupid Apple :/
| |
| [12:20 PM PT] Cuga.Rajal @rajal.org:9000: yvw :) glad it happened after I was back from travel
| |
| [12:22 PM PT] Gavin.Hird @grid.xmir.org:8002: if I do codesign -v -v on your old libs
| |
| [12:23 PM PT] Gavin.Hird @grid.xmir.org:8002: it sayus:
| |
| [12:23 PM PT] Gavin.Hird @grid.xmir.org:8002: libBulletSim.dylib: valid on disk
| |
| [12:23 PM PT] Ubit Umarov: those that where testing fs 7.0 alpha and did update, somethings no longer work due to ll change on protocol
| |
| [12:23 PM PT] Gavin.Hird @grid.xmir.org:8002: libBulletSim.dylib: satisfies its Designated Requirement
| |
| [12:23 PM PT] Cuga.Rajal @rajal.org:9000: try codesign -d --verbose=2 --extract-certificates [file]
| |
| [12:23 PM PT] Ubit Umarov: changes on inveotry procotol seems working on that and main beta
| |
| [12:24 PM PT] Ubit Umarov: just it shows now inventory thumnails and those do not work
| |
| [12:24 PM PT] Gavin.Hird @grid.xmir.org:8002: [REDACTED]
| |
| [12:24 PM PT] Ubit Umarov: since they only use code dependent on AISv3
| |
| [12:24 PM PT] Gavin.Hird @grid.xmir.org:8002: did I dox you now?
| |
| [12:25 PM PT] Ubit Umarov: wel thats a thing abotu soft sign.. no anonimate
| |
| [12:26 PM PT] Ubit Umarov: well unless apple allows it?
| |
| [12:26 PM PT] Gavin.Hird @grid.xmir.org:8002: they don't
| |
| [12:26 PM PT] Vincent.Sylvester @hg.zetaworlds.com: LL rushing things out more than usual working on different things at the same time causing everything to break huh
| |
| [12:26 PM PT] Gavin.Hird @grid.xmir.org:8002: Just edit the above out from the minutes, pls
| |
| [12:27 PM PT] Gavin.Hird @grid.xmir.org:8002: yeah, their development process is a complete mess as seen from outside
| |
| [12:27 PM PT] Ubit Umarov: Ohh vicent canm you edit out that name, later, before uploading the log?
| |
| [12:27 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Will do
| |
| [12:27 PM PT] Cuga.Rajal @rajal.org:9000: gavin is that from ther old libs? I get a different result, weird
| |
| [12:27 PM PT] Ubit Umarov: thanks
| |
| [12:27 PM PT] Gavin.Hird @grid.xmir.org:8002: yes it is
| |
| [12:28 PM PT] Cuga.Rajal @rajal.org:9000: well it's nice to see, but not clear why we are gettign diff results. Will look into this later
| |
| [12:28 PM PT] Gavin.Hird @grid.xmir.org:8002: yup
| |
| [12:30 PM PT] Joe Magarac: Yes, LL is pushing out more stuff. But there was a 3-year period during which they did very little. This is progress
| |
| [12:30 PM PT] Gavin.Hird @grid.xmir.org:8002: actually they did a heck of a lot in the background Joe
| |
| [12:30 PM PT] Andrew Hellershanks: brb
| |
| [12:30 PM PT] Ubit Umarov: well would be progress if it where good things.. but details ;)
| |
| [12:30 PM PT] Gavin.Hird @grid.xmir.org:8002: as they seem to do agaon
| |
| [12:31 PM PT] Joe Magarac: Right, they had to clean up a lot to get onto AWS. Old versions of libraries...
| |
| [12:31 PM PT] Ubit Umarov: hmm they are at aws for years now
| |
| [12:31 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I'd wish they'd think a bit more about how things get implemented so the new functions in LSL would make a bit more sense
| |
| [12:32 PM PT] Ubit Umarov: thing is that they had moved to sansar..
| |
| [12:32 PM PT] Gavin.Hird @grid.xmir.org:8002: I have said it before, I am pretty sure they are developing an xbox version of the viewer
| |
| [12:32 PM PT] Ubit Umarov: and did very little on sl
| |
| [12:33 PM PT] Gavin.Hird @grid.xmir.org:8002: they may also drop Mac support, because they have updated the build instructions for the Windows version, but the Mac version instructions are significantly outdated
| |
| [12:33 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Yeah those game control changes are annoying too, as if xbox was the only controller out there, would have liked some more modularity on that
| |
| [12:34 PM PT] Cuga.Rajal @rajal.org:9000: and yet they are working on an iPad viewer
| |
| [12:34 PM PT] Jagga Meredith: kewl
| |
| [12:34 PM PT] Ubit Umarov: those on top of the mobile viewer?
| |
| [12:34 PM PT] Joe Magarac: Game controller support will use the Human Interface Device standard. That supports most things that plug into a USB port.
| |
| [12:34 PM PT] Gavin.Hird @grid.xmir.org:8002: more serious is their chief tech (cannot remember his name) said i one of their podcasts / videos that "Mac users will lag behind Windows users in functionality and performance, but we will try to have an option also for these users"
| |
| [12:34 PM PT] Ubit Umarov: or those are realy the mobile project?
| |
| [12:34 PM PT] Cuga.Rajal @rajal.org:9000: no idea, its a hush-hush Linden project
| |
| [12:34 PM PT] Cuga.Rajal @rajal.org:9000: bhut they do post their progress from time to time
| |
| [12:35 PM PT] Joe Magarac: It's apparently written in C# but that's all I know.
| |
| [12:35 PM PT] Cuga.Rajal @rajal.org:9000: LL posted an update about it in October
| |
| [12:35 PM PT] Gavin.Hird @grid.xmir.org:8002: The mobile project is running on Vulkan
| |
| [12:35 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Mobile viewer gonna work real well with lludp and spotty connections unless they want to move all into tcp now, yay protocol overhead isn't something we already have enough of
| |
| [12:35 PM PT] Ubit Umarov: c# on xbox? ll is really good on using the wrong tools for a job ;)
| |
| [12:36 PM PT] Joe Magarac: I'm wondering if the mobile viewer has some extra server support.
| |
| [12:36 PM PT] Cuga.Rajal @rajal.org:9000: I suspect they are not doing any UDP
| |
| [12:36 PM PT] Cuga.Rajal @rajal.org:9000: but thats just a guess
| |
| [12:36 PM PT] Joe Magarac: Don't know.
| |
| [12:36 PM PT] Gavin.Hird @grid.xmir.org:8002: there is a crippled version of Vulcan called MoltenVK, that translates it to Metal, but it is a subset only
| |
| [12:36 PM PT] Gavin.Hird @grid.xmir.org:8002: and Apple has made major updates to Metal lately
| |
| [12:37 PM PT] Ubit Umarov: old idea for mobile was kinda server side video streaming of a client
| |
| [12:37 PM PT] Ubit Umarov: soemthing done for several games... that even can be played on tvs
| |
| [12:37 PM PT] Joe Magarac: There is a plan to migrate some more things to the event poller channel. But after the Materials debacle, that isn't looking as good. Too much LLSD overhead.
| |
| [12:37 PM PT] Ubit Umarov: bahh materials problem was not llsd overhaed
| |
| [12:38 PM PT] Ubit Umarov: was/is LSL nonsense
| |
| [12:38 PM PT] Gavin.Hird @grid.xmir.org:8002: I will not be surprised they are writing a DirectX 12 renderer, and dropping Mac support
| |
| [12:38 PM PT] Ubit Umarov: plus JSON embeded on llsdxml
| |
| [12:38 PM PT] Joe Magarac: Sure it was. Someone made a disco floor where every tile changed material on every beat. 50 MB/minute of LLSD over the event poller channel.
| |
| [12:39 PM PT] Ubit Umarov: gltf sees a fun thign to help sells... but a very bad format to use on fast streaming
| |
| [12:39 PM PT] Joe Magarac: Agreed.
| |
| [12:39 PM PT] Ubit Umarov: seems..
| |
| [12:39 PM PT] Joe Magarac: So t hey changed to some UDP mesages for material overrides.
| |
| [12:39 PM PT] Ubit Umarov: wo yea at least they changed that...
| |
| [12:40 PM PT] Ubit Umarov: sadly adding extra work on servers.. bc tehy still upload data in gltf and send to viewers in compact LLSD notation
| |
| [12:40 PM PT] Ubit Umarov: guess theirs servers have to do that transcode on the fly per prim face... so clever...
| |
| [12:40 PM PT] Vincent.Sylvester @hg.zetaworlds.com: This is going to strain viewer development in regards to compatibility and make every code merge even more annoying
| |
| [12:40 PM PT] Gavin.Hird @grid.xmir.org:8002: Does anyone know what the Sansar renderer was written in (it did not have Mac support)
| |
| [12:41 PM PT] Joe Magarac: No idea how Sansar did things.
| |
| [12:41 PM PT] Joe Magarac: I use WGPU, which is supposed to supprot Mac, but I havent' tested it.
| |
| [12:41 PM PT] Ubit Umarov: looking to new ll code.. seems sansar did things badly :P
| |
| [12:41 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: silly question time ~~ are we this dependant on LL?
| |
| [12:41 PM PT] Andrew Hellershanks: I'm back.
| |
| [12:41 PM PT] Gavin.Hird @grid.xmir.org:8002: I suppose one could find out if there is an old client installer somewhere
| |
| [12:41 PM PT] Joe Magarac: If a Mac dev would like to try, talk to me.
| |
| [12:41 PM PT] Gavin.Hird @grid.xmir.org:8002: yes we are
| |
| [12:42 PM PT] Gavin.Hird @grid.xmir.org:8002: OpenSim viewers are 95% LL code
| |
| [12:42 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: is that healthy?
| |
| [12:42 PM PT] Ubit Umarov: wel the dependence rope may be abotu to break( again )
| |
| [12:42 PM PT] Orbert.Tatham @hg.zetaworlds.com: It has never been "healthy"
| |
| [12:42 PM PT] Joe Magarac: Right.
| |
| [12:42 PM PT] Clifford.Hanger @alternatemetaverse.com:8002 listens
| |
| [12:42 PM PT] Ubit Umarov: some ll changes are jsut too bad to follow
| |
| [12:42 PM PT] Joe Magarac: I am behind on keeping Sharpview up with LL. Sharpview is 0% LL code. 100% Rust.
| |
| [12:43 PM PT] Orbert.Tatham @hg.zetaworlds.com: We are dependent on the viewer writers, and they are the ones that don't want to separate us
| |
| [12:43 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: understood
| |
| [12:43 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: includes dayturm?
| |
| [12:43 PM PT] Orbert.Tatham @hg.zetaworlds.com: I am hoping Joe here will give us a code base that allows the separation
| |
| [12:43 PM PT] Ubit Umarov: well when fs did sperated us, they stopped doing any code for us..
| |
| [12:43 PM PT] Gavin.Hird @grid.xmir.org:8002: the problem is that nobody are capable of rewriting the renderer
| |
| [12:43 PM PT] Gavin.Hird @grid.xmir.org:8002: yes also dayturn
| |
| [12:44 PM PT] Ubit Umarov: dayturn already does not ahve some ll blablacode :)
| |
| [12:44 PM PT] Joe Magarac: Whenevery someone wants the code, I point them at an open source test of the rendering stack that I have, and ask them to compile and build it. So far, nobody has.
| |
| [12:44 PM PT] Gavin.Hird @grid.xmir.org:8002: on macOS Apple has said they will park OpenGL and the only thing running on macOS is Metal
| |
| [12:44 PM PT] Gavin.Hird @grid.xmir.org:8002: once OpenGL is toast
| |
| [12:45 PM PT] Ubit Umarov: seem thing thing vulkan will run on mac??
| |
| [12:45 PM PT] Ubit Umarov: ...think...
| |
| [12:45 PM PT] Gavin.Hird @grid.xmir.org:8002: it wont
| |
| [12:45 PM PT] Gavin.Hird @grid.xmir.org:8002: it is a subset
| |
| [12:45 PM PT] Joe Magarac: If you want to try, build this: https://github.com/John-Nagle/ui-mock
| |
| [12:45 PM PT] Gavin.Hird @grid.xmir.org:8002: and Apple has done major changes in Metal 3 to make support for the VR headset
| |
| [12:45 PM PT] Gavin.Hird @grid.xmir.org:8002: that Vulkan cannot do
| |
| [12:46 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It's multiple things combined that make it so bad. The practice of merging mainline code then reverting and changing things back to add compatibility or fix LL code. Then you have a large dependency tree and quite complex setup to build the viewer making development slow as you wait for it to build. So it's a dev and devops issue and you still have to watch out for LL curve balls
| |
| [12:46 PM PT] Joe Magarac: It's a dummy game - same stack as Sharpview.
| |
| [12:46 PM PT] Ubit Umarov: "100% safe Rust." hmm :)
| |
| [12:46 PM PT] Gavin.Hird @grid.xmir.org:8002: correct Vincent
| |
| [12:46 PM PT] Joe Magarac: Works on windows and Linux. Try building it on MacOS and let me know how that works.
| |
| [12:46 PM PT] Gavin.Hird @grid.xmir.org:8002: it is a real pain
| |
| [12:46 PM PT] Ubit Umarov: i don't think my car rust is 100% safe ..
| |
| [12:47 PM PT] Gavin.Hird @grid.xmir.org:8002: bookmarked your repo
| |
| [12:47 PM PT] Joe Magarac: Try that out, please. If there are problems I will help. But I don't have any Mac gear.
| |
| [12:48 PM PT] Ubit Umarov: guess rust like c# will not need sign on mac?
| |
| [12:48 PM PT] Ubit Umarov: wel assuming c# will not need it..
| |
| [12:48 PM PT] Joe Magarac: It produces hard executables, so it will probably need signing.
| |
| [12:49 PM PT] Ubit Umarov: yeah was thinking that on AOT compile
| |
| [12:49 PM PT] Gavin.Hird @grid.xmir.org:8002: even if it is running Rust, in will have to be compiled in Xcode for the final app to run as a desktop app, so ideally it must be singed and notraized
| |
| [12:49 PM PT] Ubit Umarov: wel in fact apple only allows AOT
| |
| [12:50 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Changing the development pipeline and cleaning up the dependency mess would not just benefit us, but make it generally easier to develop a viewer using LL code. That's a big undertaking though and it doesn't completely solve the compatibility problem either
| |
| [12:50 PM PT] Gavin.Hird @grid.xmir.org:8002: dotnet c# like opensium must not be singed
| |
| [12:50 PM PT] Gavin.Hird @grid.xmir.org:8002: signed
| |
| [12:50 PM PT] Andrew Hellershanks: Joe, What other dependencies are there to build the ui-mock apart from needing the rust dev tools?
| |
| [12:50 PM PT] Gavin.Hird @grid.xmir.org:8002: xcode isn't involved at all in dotnet builds
| |
| [12:50 PM PT] Joe Magarac: It's all handled by Rust's Cargo system.
| |
| [12:50 PM PT] Ubit Umarov: opensim could be pre compiled also
| |
| [12:50 PM PT] Ubit Umarov: in AOT
| |
| [12:51 PM PT] Ubit Umarov: ms starts to "sell" that idea..
| |
| [12:51 PM PT] Joe Magarac: Just install Rust, try "cargo build", and see what happens.
| |
| [12:51 PM PT] Andrew Hellershanks: As a check the rust-all package pulls in 12 other packages.
| |
| [12:51 PM PT] Ubit Umarov: since JIT is... well damm slow :)
| |
| [12:51 PM PT] Gavin.Hird @grid.xmir.org:8002: sure Ubit, but it does not run as a desktop app
| |
| [12:51 PM PT] Gavin.Hird @grid.xmir.org:8002: it is invoked from the BSD subsystem, and does not need to be signed to run
| |
| [12:51 PM PT] Ubit Umarov: well i stest AOT long ago still on mono
| |
| [12:51 PM PT] Orbert.Tatham @hg.zetaworlds.com: Andrew, probably the best way to install it is to use the rust script on their web site
| |
| [12:52 PM PT] Ubit Umarov: not the new dotnet6 forms
| |
| [12:52 PM PT] Ubit Umarov: hmm or dotnet8
| |
| [12:52 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I think we ultimately concluded compiling in release vs debug mode made a bigger difference than aot for most things
| |
| [12:52 PM PT] Gavin.Hird @grid.xmir.org:8002: I suppose I could try and compile opensim in VS for macOS
| |
| [12:52 PM PT] Gavin.Hird @grid.xmir.org:8002: but I cannot see it makes a difference
| |
| [12:53 PM PT] Ubit Umarov: well performance change btw AOT and JIT is mostly on methods loading
| |
| [12:53 PM PT] Gavin.Hird @grid.xmir.org:8002: if it was an iOS app with na UI, sure, then i had to be signed, and the final build is done by Xcode
| |
| [12:53 PM PT] Gavin.Hird @grid.xmir.org:8002: but not for something invoked by the BSD subsystem
| |
| [12:53 PM PT] Ubit Umarov: once a method is compiled into native, things stay similar.. in fact jit maybe a bit better
| |
| [12:55 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: ok 5 mins left - Is there any work/progress on voice in the last 2 weeks?
| |
| [12:55 PM PT] Gavin.Hird @grid.xmir.org:8002: nothing but I have freed up a machine to try the mac version on Mumble
| |
| [12:55 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Rome wasn't built in a day :)
| |
| [12:55 PM PT] Gavin.Hird @grid.xmir.org:8002: and downloaded the executable
| |
| [12:55 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: lol vincent
| |
| [12:56 PM PT] Vincent.Sylvester @hg.zetaworlds.com: One of the things that I can't help but notice is that the increase in performance in dotnet does come along with higher cpu usage, which makes me wonder if some code is now so efficient it effectively spams now
| |
| [12:57 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Something running twice as fast without any throttle could cause bottlenecks elsewhere
| |
| [12:57 PM PT] Cuga.Rajal @rajal.org:9000: CPU is always underutilized so thats a good thing
| |
| [12:57 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: not seen any dramatic increases
| |
| [12:58 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: what i have seen is the opensim library has dissapeared
| |
| [12:58 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: but that may be a config issue
| |
| [12:58 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: on my part
| |
| [12:59 PM PT] Ubit Umarov: what opensim lib?
| |
| [12:59 PM PT] Andrew Hellershanks: Vincent, Any idea how much of an increase there was in CPU usage?
| |
| [12:59 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: in the viewer inventory
| |
| [12:59 PM PT] Ubit Umarov: well we do have it..
| |
| [1:00 PM PT] Ubit Umarov: even on fs beta with recent ll changes to inventory ( guess dayturn does not have them still )
| |
| [1:00 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: must be my robust split, missing something
| |
| [1:00 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Probably somewhere in the range of 15-30% depending on case. I did notice that it used to be somewhat linear up until a point and now it tends to go up in usage much more quickly and then slows down a bit
| |
| [1:00 PM PT] Orbert.Tatham @hg.zetaworlds.com: Does the RAM usages match that?
| |
| [1:01 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It's scary to see usage of 250% cpu with a lot of users
| |
| [1:01 PM PT] Andrew Hellershanks: Vincent, that's not going to be good for someone running a lot regions on a single box.
| |
| [1:01 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Memory usage is way down compared to say three years ago
| |
| [1:01 PM PT] Orbert.Tatham @hg.zetaworlds.com: Probably part of the tradeoff
| |
| [1:01 PM PT] Ubit Umarov: i did made several changes to how we did use memory | | [1:01 PM PT] Ubit Umarov: i did made several changes to how we did use memory |
| [1:01 PM PT] Ubit Umarov: and dotnet gc is a bit better | | [1:01 PM PT] Ubit Umarov: and dotnet gc is a bit better |
| [1:02 PM PT] Lyr Lobo grins and nods
| |
| [1:02 PM PT] Ubit Umarov: but for example all lludp packet buffers are now reused | | [1:02 PM PT] Ubit Umarov: but for example all lludp packet buffers are now reused |
| [1:02 PM PT] Ubit Umarov: that is a lot less stress on gc | | [1:02 PM PT] Ubit Umarov: that is a lot less stress on gc |
| Ligne 284 : |
Ligne 69 : |
| [1:04 PM PT] Ubit Umarov: lludp sending is now also done with fastfoward serialization without intermediate structures | | [1:04 PM PT] Ubit Umarov: lludp sending is now also done with fastfoward serialization without intermediate structures |
| [1:05 PM PT] Ubit Umarov: and some some caps.. | | [1:05 PM PT] Ubit Umarov: and some some caps.. |
| [1:05 PM PT] Joe Magarac: Next meeting, bye for now.
| |
| [1:05 PM PT] Jagga Meredith: could that be within the real of statistical error? | | [1:05 PM PT] Jagga Meredith: could that be within the real of statistical error? |
| [1:05 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: bye joe
| |
| [1:05 PM PT] Andrew Hellershanks: ok, Joe. Thanks for dropping by.
| |
| [1:05 PM PT] Jagga Meredith: realm | | [1:05 PM PT] Jagga Meredith: realm |
| [1:05 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: i think so | | [1:05 PM PT] Clifford.Hanger @alternatemetaverse.com:8002: i think so |
| Ligne 295 : |
Ligne 77 : |
| [1:06 PM PT] Ubit Umarov: well cpu load is not that a big metric... things need to use cpu to work :P | | [1:06 PM PT] Ubit Umarov: well cpu load is not that a big metric... things need to use cpu to work :P |
| [1:07 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It's difficult to measure that stuff accurately given the amount of variables you have in play. It is mostly just looking at usage during peak hours and comparing that to how things used to behave. Noticing mostly a difference in how usage grows more quickly and then levels off, over a more linear growth | | [1:07 PM PT] Vincent.Sylvester @hg.zetaworlds.com: It's difficult to measure that stuff accurately given the amount of variables you have in play. It is mostly just looking at usage during peak hours and comparing that to how things used to behave. Noticing mostly a difference in how usage grows more quickly and then levels off, over a more linear growth |
| [1:16 PM PT] Ubit Umarov: bad network weather
| |
| [1:18 PM PT] Lyr Lobo: bye Vincent!
| |
| [1:18 PM PT] Ubit Umarov: ok fun networkj glitch
| |
| [1:18 PM PT] Lyr Lobo: sorry to dash so fast *laughs* yes, pulled the plug, eh?
| |
| [1:18 PM PT] Lyr Lobo teases
| |
| [1:19 PM PT] Ubit Umarov: guess someone on a major net node pulled out the wrong cable
| |
| [1:19 PM PT] Vincent.Sylvester @hg.zetaworlds.com: One way to end a meeting
| |
| [1:19 PM PT] Lyr Lobo: hehe
| |
| [1:19 PM PT] Lyr Lobo: see ya!
| |
| [1:19 PM PT] Ubit Umarov: :)
| |
| [1:20 PM PT] Ubit Umarov: yeah rl also calls me :)
| |
| [1:20 PM PT] Ubit Umarov: cya aroung
| |
| [1:20 PM PT] Vincent.Sylvester @hg.zetaworlds.com: The logging...
| |
| [1:21 PM PT] Andrew Hellershanks: That was an unexpected end to the meeting.
| |
| [1:22 PM PT] Andrew Hellershanks: We all got logged out and the region went away for a bit but we are back.
| |
| [1:22 PM PT] Vincent.Sylvester @hg.zetaworlds.com: mobile viewer experience lol
| |
| [1:22 PM PT] Andrew Hellershanks: As the meeting was just about over I will officially mark it as ended and turn off the chat logger.
| |
| --> | | --> |