Réunion du 11-01-2022

Source :http://opensimulator.org/wiki/Chat_log_from_the_meeting_on_2022-01-11

Introduction

• Migration des assets sur Osgrid
• Viewer RLV

[11:00] Kayaker Magic : Et moi qui espérais que la réunion soit courte....
[11:00] Gavin.Hird @grid.xmir.org:8002 : Bonjour Andrew.
[11:00] Ubit Umarov : oh et il y a le andrew le banc 
[11:00] Ubit Umarov : blanc
[11:00] Andrew Hellershanks : Bonjour, tout le monde.
[11:00] Andrew Hellershanks : Je peux faire en sorte que la réunion soit courte.
[11:00] Andrew Hellershanks : Merci à tous d'être venus. Nous nous reverrons la semaine prochaine.
[11:00] Andrew Hellershanks sourit.
[11:00] JayR.Cela @grid.3rdrockgrid.com:8002: lol
[11:01] Ubit Umarov : Les assets d'osgrid pourraient prendre encore 120 jours.
[11:01] Ubit Umarov: gezz
[11:01] Andrew Hellershanks : Pourquoi une réunion courte, Kayaker.
[11:01] Andrew Hellershanks : 120 jours ? Eek.
[11:01] Andrew Hellershanks : Je vais être un Andrew le blanc pendant un bon moment encore.
[11:01] Gavin.Hird @grid.xmir.org:8002 : Mon dieu, sur quelle variante de covid nous serons à ce moment là ?
[11:01] Kayaker Magic : Ubit plaisantait en disant au revoir, à la semaine prochaine.
[11:01] Jagga Meredith: zulu
[11:01] Ubit Umarov: un potentiel Andrew le rose aurait été pire.
[11:02] Andrew Hellershanks : Je vois mon avatar en gris.
[11:02] Gavin.Hird @grid.xmir.org:8002 : espace gris.
[11:02] Kayaker Magic : Andrew : tu es blanc pour moi, pas de textures.
[11:03] Andrew Hellershanks : Eh bien, le gris est un blanc cassé, en quelque sorte.
[11:03] Andrew Hellershanks : :)
[11:03] Jagga Meredith : Je vais bien.  Mais je n'ai pas changé de tenue depuis environ 5 ans.
[11:03] Gavin.Hird @grid.xmir.org:8002 : il est probablement blanc, mais l'éclairage le fait paraître gris clair.
[11:03] Andrew Hellershanks : Je n'ai pas changé de tenue depuis longtemps non plus.
[11:03] Ubit Umarov : oui, il y a des nuances de gris ici aussi.
[11:03] Gavin.Hird @grid.xmir.org:8002 : Je pense qu'aucun d'entre nous ne l'a fait, haha.
[11:04] Andrew Hellershanks murmure : Je semble être le "chanceux" car je ne vois personne d'autre avec un avatar gris ou blanc. Devrais-je acheter un billet de loterie ?
[11:04] Ubit Umarov : ( et plus de cinquante)
[11:04] Gavin.Hird @grid.xmir.org:8002 : mais c'est seulement si tu utilises le viewer RLV Ubit.
[11:04] Jamie.Jordan @grid.kitely.com:8002 : Bonjour tout le monde.
[11:05] Andrew Hellershanks : Ubit a cessé de porter son masque Covid.
[11:05] Gavin.Hird @grid.xmir.org:8002 : il a abandonné.
[11:05] Andrew Hellershanks : Bonjour, Jamie
[11:05] Gavin.Hird @grid.xmir.org:8002 : Salut Jamie
[11:05] Selby.Evans @grid.kitely.com:8002 : Salut Jamie
[11:05] Gavin.Hird @grid.xmir.org:8002 : ... ou l'inventaire est en cours de migration.
[11:07] Jagga Meredith : à quoi servait la panne du week-end ?
[11:08] Ubit Umarov : oops masque ??
[11:08] Ubit Umarov : Ohh noo
[11:08] Andrew Hellershanks : Ubit a été occupé la semaine dernière. J'ai touché à beaucoup de fichiers afin d'examiner la façon dont certains codes étaient générés. Cela peut aider un peu à améliorer les performances du code de la grille.
[11:08] Ubit Umarov : corrigé

Le service d'assets de OSGrid : code privé et fermé

• MariaDB
• MySQL
• Lbsa Plaza sur OSGrid

[11:09] Ubit Umarov : pas d'information officielle de la part d'osgrid.
[11:10] Ubit Umarov : mais il semble que mariadb fonctionnant en mode cluster / maître -esclave vient d'exploser à nouveau.
[11:10] Ubit Umarov : et remplacé maintenant par un autre fork de mysql
[11:10] Andrew Hellershanks : ouch
[11:10] Gavin.Hird @grid.xmir.org:8002 : il est peut-être temps d'utiliser une vraie base de données.
[11:11] Gavin.Hird @grid.xmir.org:8002 : fork de mysql ?
[11:11] Ubit Umarov : oui.
[11:11] Gavin.Hird @grid.xmir.org:8002 : qui serait ?
[11:11] Ubit Umarov : ( comme l'était Mariadb )
[11:11] Ubit Umarov : je n'ai pas de déclarations officielles sur osgrid :)
[11:12] Gavin.Hird @grid.xmir.org:8002 : Je suppose qu'ils ne maintiennent pas le fork
[11:12] Gavin.Hird @grid.xmir.org:8002 : Je me demande quel est le nom du projet forké.
[11:12] Ubit Umarov : ces services d'assets ne sont même pas du code opensimulator.
[11:12] Ubit Umarov : mais une solution privée fermée... donc... nous ne pouvons pas en dire grand chose.
[11:12] Gavin.Hird @grid.xmir.org:8002 : alors est-ce que c'est encore une grille de test ?
[11:13] Ubit Umarov : oui test possible pour cette solution fermée et privée, pas pour opensimulator, dans ce domaine :)
[11:13] Ubit Umarov : mais lbsa est en train de brûler le (presque) dernier code.
[11:13] Ubit Umarov : alors cette région est-elle en train de brûler ?
[11:13] Andrew Hellershanks : brûle-t-elle ? :)
[11:14] Ubit Umarov : en fait, c'est la dernière.

Brûler vs Graver : humour de dev

Ndlr : j'avoue ne pas avoir tout compris, la traduction peut-être très approximative. :)

• RS/600
• AFAIK : As Far As I Know = Pour autant que je sache.

[11:14] Gavin.Hird @grid.xmir.org:8002 : brûler des cycles de CPU
[11:14] Ubit Umarov : ouais :)
[11:14] Andrew Hellershanks : Oui.
[11:14] Ubit Umarov : si je me souviens bien, à l'époque de la production des circuits intégrés, il y avait une étape optionnelle de gravure.
[11:15] Gavin.Hird @grid.xmir.org:8002 : brûlage ou burn-in ?
[11:15] Ubit Umarov : quand les circuits intégrés étaient placés sur un circuit fonctionnant pendant plusieurs heures.
[11:15] Andrew Hellershanks: burn-in (graver)
[11:15] Ubit Umarov : ou cela
[11:16] Andrew Hellershanks : Ce serait brûlé si le test échoue ;)
[11:16] Gavin.Hird @grid.xmir.org:8002 : :-)
[11:16] Ubit Umarov : le problème est que la probabilité d'un défaut a une forme de baignoire (bath tube).
[11:16] Gavin.Hird @grid.xmir.org:8002 : le RS/6000 avait un code de console HW qui disait "le CPU est en feu".
[11:16] Ubit Umarov : élevé dans les premières heures de la vie d'une pièce... puis élevé à nouveau en fin de vie.
[11:17] Ubit Umarov : donc le fait traiter la pièce avant la production a permis d'exclure un plus grand nombre de ces échecs précoces.
[11:17] Andrew Hellershanks : Je me souviens qu'un produit développé par une entreprise pour laquelle je travaillais il y a quelques 
années mettait les composants dans une pièce pour les brûler à haute température.
[11:18] Ubit Umarov : donc pour le coup... code brûlant à lbsa ;)
[11:18] Gavin.Hird @grid.xmir.org:8002 : AFAIK tous les appareils Apple sont soumis à un burn-in avant d'être envoyés aux clients.
[11:19] Andrew Hellershanks : Je dirais que c'est une pratique courante dans l'industrie de faire un burn-in du produit.
[11:19] Gavin.Hird @grid.xmir.org:8002: oui
[11:19] Ubit Umarov : pas vraiment.
[11:19] Ubit Umarov : sur les produits de consommation bon marché.
[11:19] Ubit Umarov : autrefois... plus maintenant.
[11:20] Ubit Umarov : cela reste possible sur les produits haut de gamme.
[11:20] Ubit Umarov : les pièces qui défaillantes sont réutilisées sur des produits moins chers ( je pense :p )
[11:20] Ubit Umarov : bien mais à propos d'opensim
[11:20] Ubit Umarov : quelles nouvelles avez-vous ?

Nouvelles d'OpenSim

[11:21] Gavin.Hird @grid.xmir.org:8002 : LL a mentionné qu'ils allaient lancer de nouvelles fonctionnalités cette année
[11:22] Andrew Hellershanks : En plus des modifications pour gagner du temps lors de tests ou de comparaisons d'UUID, il y a eu une ou plusieurs mises à jour de la bibliothèque libomv.
[11:22] Ubit Umarov: telling otherwise would mean "we are dead" :)
[11:22] Gavin.Hird @grid.xmir.org:8002: so keep your loins girded
[11:22] Ubit Umarov: changes to libomv where related
[11:22] Ubit Umarov: bc uuid code is at libomv and was change
[11:22] Ubit Umarov: changed
[11:22] Andrew Hellershanks: I thought that was the case by the commit messages.
[11:23] Ubit Umarov: so a lot of files changed
[11:24] Andrew Hellershanks: In reviewing Ubits changes I only noted about 3 or so other changes to be made but no errors. Took quite a while to review the commits as they hit so many files.
[11:24] Ubit Umarov: with changes like  auuid == UUID.Zero replaced by a new auuis.IsZero()
[11:24] Ubit Umarov: for example
[11:24] Andrew Hellershanks: I wonder what diffstat would report as to the count of changes.
[11:24] Ubit Umarov: first form means at least 2 blocks of 16bytes copies
[11:25] Ubit Umarov: as part of argument passing to the function that does the ==
[11:25] Ubit Umarov: in real JIT code, and bc there where a chain of calls... there where like 8 copies
[11:25] Ubit Umarov: all useless.. but well
[11:26] Andrew Hellershanks: It seems a little odd that it would need to copy data just to do a comparison.
[11:26] Ubit Umarov: auuid.IsZero on the other is a no arguments call to a methog that does check of zero
[11:26] Ubit Umarov: so a lot faster
[11:26] Andrew Hellershanks: I would have thought it could have done the comparison in place using the original variable location.
[11:27] Ubit Umarov: yeah Trust on compilers :P
[11:27] Andrew Hellershanks: yea. It can be quite revealing to look at the generated assembler code.
[11:27] Gavin.Hird @grid.xmir.org:8002: by copying it to both sides, the test will always be true
[11:27] Ubit Umarov: and != could even have another set of copies
[11:28] Ubit Umarov: for example
[11:28] Ubit Umarov: if(ua != ub)  was
[11:28] Ubit Umarov: <Ubit_> <Ubit_> 00007FFB371C10FB  vmovdqu     xmm0,xmmword ptr [rsp+200h]  
<Ubit_> <Ubit_> 00007FFB371C1105  vmovdqu     xmmword ptr [rsp+118h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C110F  vmovdqu     xmm0,xmmword ptr [rsp+210h]  
<Ubit_> <Ubit_> 00007FFB371C1119  vmovdqu     xmmword ptr [rsp+108h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C1123  vmovdqu     xmm0,xmmword ptr [rsp+118h]  
<Ubit_> <Ubit_> 00007FFB371C112D  vmovdqu     xmmword ptr [rsp+0F8h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C1137  vmovdqu     xmm0,xmmword ptr [rsp+108h]  
<Ubit_> <Ubit_> 00007FFB371C1141  vmovdqu     xmmword ptr [rsp+0E8h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C114B  vmovdqu     xmm0,xmmword ptr [rsp+0F8h]  
<Ubit_> <Ubit_> 00007FFB371C1155  vmovdqu     xmmword ptr [rsp+38h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C115C  vmovdqu     xmm0,xmmword ptr [rsp+0E8h]  
<Ubit_> <Ubit_> 00007FFB371C1166  vmovdqu     xmmword ptr [rsp+28h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C116D  lea         rcx,[rsp+38h]  
<Ubit_> <Ubit_> 00007FFB371C
[11:28] Ubit Umarov:  rdx,[rsp+28h]  
<Ubit_> <Ubit_> 00007FFB371C1177  call        System.Guid.op_Equality(System.Guid, System.Guid) (07FFB8FD4BE90h)  
<Ubit_> <Ubit_> 00007FFB371C117C  test        al,al  
<Ubit_> <Ubit_> 00007FFB371C117E  jne         testlixo.Program.Main(System.String[])+03E2h (
[11:29] Ubit Umarov: in windows release mode, i7 native code
[11:30] Andrew Hellershanks: What was the code doing previously for the zero check?
[11:31] Andrew Hellershanks: While Ubit chews on that question, any other questions about OpenSim?
[11:32] JayR.Cela @grid.3rdrockgrid.com:8002: none from me
[11:32] Andrew Hellershanks looks at Kayaker
[11:32] Ubit Umarov: was same
[11:33] Andrew Hellershanks: Hello, Michael.
[11:33] Bill's Chair whispers: This is Bill's Chair
[11:33] Gavin.Hird @grid.xmir.org:8002: Hi Michael
[11:33] Michael.Christopher @cc.opensimulator.org:8002: hello!
[11:33] Ubit Umarov: UUID.Zero is a uuid like other, taking place of ub there
[11:33] Andrew Hellershanks: You can't sit there, Michael. Pick another seat.
[11:33] Ubit Umarov: so same code
[11:33] Andrew Hellershanks: Right.
[11:34] Kayaker Magic: I'm good this week
[11:34] Andrew Hellershanks: ok
[11:34] Andrew Hellershanks: Anyone else?
[11:34] Michael.Christopher @cc.opensimulator.org:8002: nothing new here this week, still waiting on the replacement developer system to arrive :/
[11:35] Ubit Umarov: now:
[11:35] Ubit Umarov: 118:             if (ua.IsZero())
00007FFB371B3317  cmp         qword ptr [rsp+190h],0  
00007FFB371B3320  jne         00007FFB371B332F  
00007FFB371B3322  cmp         qword ptr [rsp+198h],0  
00007FFB371B332B  jne         00007FFB371B332F
[11:35] Andrew Hellershanks: Santa is a bit slow on his delivery of the system, Michael?  ;)
[11:35] Ubit Umarov: just compares 4 ulongs
[11:35] Ubit Umarov: and direct, and inline.. no copies
[11:36] Michael.Christopher @cc.opensimulator.org:8002: lol yea.  We ordered back in December and it just went to "shipping" status today *eye roll*
[11:36] Gavin.Hird @grid.xmir.org:8002: right, try purchase a 8GB Raspberry PI somewhere
[11:36] Gavin.Hird @grid.xmir.org:8002: you can get them at 3x normal price
[11:36] Ubit Umarov: also now
[11:36] Ubit Umarov: 115:             if (ua != ub)
00007FFB371B32C9  vmovdqu     xmm0,xmmword ptr [rsp+190h]  
00007FFB371B32D3  vmovdqu     xmmword ptr [rsp+0E8h],xmm0  
00007FFB371B32DD  vmovdqu     xmm0,xmmword ptr [rsp+1A0h]  
00007FFB371B32E7  vmovdqu     xmmword ptr [rsp+0D8h],xmm0  
00007FFB371B32F1  mov         rcx,qword ptr [rsp+0E8h]  
00007FFB371B32F9  cmp         rcx,qword ptr [rsp+0D8h]  
00007FFB371B3301  jne         00007FFB371B3315  
00007FFB371B3303  mov         rcx,qword ptr [rsp+0F0h]  
00007FFB371B330B  cmp         rcx,qword ptr [rsp+0E0h]  
00007FFB371B3313  je          00007FFB371B3317
[11:36] Andrew Hellershanks: I just have my 4G Pi 4.
[11:36] Ubit Umarov: just one copy
[11:36] Gavin.Hird @grid.xmir.org:8002: the 4GB is also out of stock all over
[11:36] Ubit Umarov: and the compare is now compare fo 2 ulongs also inline
[11:37] Gavin.Hird @grid.xmir.org:8002: Canon are asking their customers to break printer toner DRM as they cannot get the chips for the toners anymore
[11:37] Ubit Umarov: vmovdqu is a intel AVX (sse) intruction that moves 16 bytes ( a a xmm register )
[11:38] Ubit Umarov: well as i said in other commits "save some CO2"  "useless changes"
[11:38] Ubit Umarov: bc we basicly see no effects :p
[11:38] Andrew Hellershanks: I wouldn't call the changes useless.
[11:38] Gavin.Hird @grid.xmir.org:8002: how about mono - I guess the asemly generated by mono is rather different
[11:38] Ubit Umarov: normal inwold users don't notice
[11:38] Gavin.Hird @grid.xmir.org:8002: assembly*
[11:39] Ubit Umarov: yeah mono JIT is dif
[11:39] Ubit Umarov: and can even be the ms one on the llvm one
[11:39] Andrew Hellershanks: No, normal users wouldn't but the people running a grid will notice the improvement.
[11:39] Ubit Umarov: yeha hope this means less overall cpu load
[11:39] Gavin.Hird @grid.xmir.org:8002: not to mention if mono also use LLVM
[11:39] Ubit Umarov: llvm is optional
[11:40] Gavin.Hird @grid.xmir.org:8002: mono on macOS is llvm standard
[11:40] Ubit Umarov: same say that ms jit is now faster.. some dont
[11:40] Gavin.Hird @grid.xmir.org:8002: which makes it a bit faster
[11:40] Ubit Umarov: wlel that past i did was a big JIT optimization fail
[11:40] Ubit Umarov: all those copies are useless
[11:41] Gavin.Hird @grid.xmir.org:8002: hopefully llvm strips the mess out
[11:41] Andrew Hellershanks: It is hard to get usable stats on the improvement because even an "idle" region still has bursts of activity. I think the change might have dropped the CPU load for my standalone by about 1% but the numbers were bouncing  around.
[11:41] Ubit Umarov: even on the new compare
[11:41] Ubit Umarov: <Ubit_> <Ubit_> 00007FFB371C10FB  vmovdqu     xmm0,xmmword ptr [rsp+200h]  
<Ubit_> <Ubit_> 00007FFB371C1105  vmovdqu     xmmword ptr [rsp+118h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C110F  vmovdqu     xmm0,xmmword ptr [rsp+210h]  
<Ubit_> <Ubit_> 00007FFB371C1119  vmovdqu     xmmword ptr [rsp+108h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C1123  vmovdqu     xmm0,xmmword ptr [rsp+118h]  
<Ubit_> <Ubit_> 00007FFB371C112D  vmovdqu     xmmword ptr [rsp+0F8h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C1137  vmovdqu     xmm0,xmmword ptr [rsp+108h]  
<Ubit_> <Ubit_> 00007FFB371C1141  vmovdqu     xmmword ptr [rsp+0E8h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C114B  vmovdqu     xmm0,xmmword ptr [rsp+0F8h]  
<Ubit_> <Ubit_> 00007FFB371C1155  vmovdqu     xmmword ptr [rsp+38h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C115C  vmovdqu     xmm0,xmmword ptr [rsp+0E8h]  
<Ubit_> <Ubit_> 00007FFB371C1166  vmovdqu     xmmword ptr [rsp+28h],xmm0  
<Ubit_> <Ubit_> 00007FFB371C116D  lea         rcx,[rsp+38h]  
<Ubit_> <Ubit_> 00007FFB371C
[11:41] Ubit Umarov:  rdx,[rsp+28h]  
<Ubit_> <Ubit_> 00007FFB371C1177  call        System.Guid.op_Equality(System.Guid, System.Guid) (07FFB8FD4BE90h)  
<Ubit_> <Ubit_> 00007FFB371C117C  test        al,al  
<Ubit_> <Ubit_> 00007FFB371C117E  jne         testlixo.Program.Main(System.String[])+03E2h (
[11:41] Ubit Umarov: those copies are also useless
[11:41] Andrew Hellershanks: wb, Jamie
[11:41] Ubit Umarov: but per spec
[11:41] Ubit Umarov: oops this is the old
[11:42] Ubit Umarov: 115:             if (ua != ub)
00007FFB371B32C9  vmovdqu     xmm0,xmmword ptr [rsp+190h]  
00007FFB371B32D3  vmovdqu     xmmword ptr [rsp+0E8h],xmm0  
00007FFB371B32DD  vmovdqu     xmm0,xmmword ptr [rsp+1A0h]  
00007FFB371B32E7  vmovdqu     xmmword ptr [rsp+0D8h],xmm0  
00007FFB371B32F1  mov         rcx,qword ptr [rsp+0E8h]  
00007FFB371B32F9  cmp         rcx,qword ptr [rsp+0D8h]  
00007FFB371B3301  jne         00007FFB371B3315  
00007FFB371B3303  mov         rcx,qword ptr [rsp+0F0h]  
00007FFB371B330B  cmp         rcx,qword ptr [rsp+0E0h]  
00007FFB371B3313  je          00007FFB371B3317
[11:43] Ubit Umarov: this is also stupid JIT code..  better but stupid
[11:43] Jamie.Jordan @grid.kitely.com:8002: thanks i'm crashy for some reason
[11:43] Ubit Umarov: JIT detected theat was no need to call a funtion
[11:43] Ubit Umarov: but failed to see that the respecitve arguments copies are also not needed
[11:44] Ubit Umarov: well  i just lauch when clever ppl tells one should trust compilers
[11:44] Ubit Umarov: laught also
[11:45] Gavin.Hird @grid.xmir.org:8002: yeah, it was better in the old days when you could hit halt, correct the code in core and continue
[11:45] Ubit Umarov: also == is bad
[11:46] Ubit Umarov: means 2 arguments
[11:46] Ubit Umarov: while auuid.Equals(other) is only 1 argument
[11:47] Ubit Umarov: so yeahh compilers are clever and see all things for us... they say
[11:47] Ubit Umarov: LOL
[11:47] Ubit Umarov: whatver
[11:47] Ubit Umarov: changed the low level code on libopenmetaverse
[11:47] Andrew Hellershanks: I had a problem one time with some code. When I dug in to the generated assembler I found that the compiler was messing around with the \n character in a string. It should have left it alone and it would be translated in to the characters used by the operating system during the output of the string. The compiler maker said they followed the ANSI standard. Handling of \n certainly wasn't being done per K&R.
[11:47] Ubit Umarov: and them on a ton of places where we use it
[11:48] Ubit Umarov: yeah problems with compilers are as old as compilets
[11:48] Ubit Umarov: gcc was very fun and still  is
[11:48] Ubit Umarov: very (NOT)  fun
[11:49] Ubit Umarov: this at least do produce correct results!!
[11:49] Ubit Umarov: and JIT was cleave to see the cpu cna copy 16bytes on a single instruction
[11:49] Ubit Umarov: not bas
[11:49] Ubit Umarov: not bad
[11:50] Ubit Umarov: except my typos
[11:50] Ubit Umarov: "and JIT was clever to see tha the  cpu can copy 16bytes on a single instruction"
[11:50] Andrew Hellershanks: :)
[11:50] Ubit Umarov: THAT nor tha
[11:50] Ubit Umarov: uff
[11:50] Ubit Umarov: get a UbitTranslator
[11:50] Ubit Umarov: :p
[11:51] Andrew Hellershanks: naww... most, if not all, of us here can read typonese.
[11:51] Ubit Umarov: :)
[11:51] Andrew Hellershanks: How many times is that, Jamie?
[11:51] Michael.Christopher @cc.opensimulator.org:8002: dont most of us writ in typonese ;)
[11:51] Gavin.Hird @grid.xmir.org:8002: now he is treading water, so he has to relog to clear that
[11:52] Ubit Umarov: i did enforce max og 64k on osmakenotecard also
[11:52] Ubit Umarov: "revert the slowdown of osMessageObject. Also reduce the slowdown osMessageAttachments to (100 + 10 *NumberTargetedPrims) miliseconds. Some action games shown a bit of visible negative impact, so lets test
"
[11:52] Andrew Hellershanks: A reasonable limit, Ubit. 64k of text is a lot of text to read. Most wouldn't read an NC with that much text in it.
[11:53] Ubit Umarov: a game here in osg had visible negative impact
[11:53] Gavin.Hird @grid.xmir.org:8002: ok?
[11:53] Ubit Umarov: "* proper bypass lludp header extra bytes, even if not seen so far
"
[11:53] Ubit Umarov: this is more of a useless change
[11:54] Ubit Umarov: happens that LLUDP protocol suports longe packets headers
[11:54] Ubit Umarov: out code would blowup if that happened
[11:54] Ubit Umarov: now will just over the extra bytes if present
[11:55] Ubit Umarov: well never seen a viewer and any use of such bytes
[11:55] Ubit Umarov: but it is a formal protocol fix
[11:55] Andrew Hellershanks: Jamie just went *poof* again.
[11:56] Andrew Hellershanks: We are just a few minutes away from the top of the hour. Any last minute questions or comments?
[11:56] Ubit Umarov: guess that covers last week code changes
[11:56] Andrew Hellershanks: Anything viewer related, Gavin?
[11:56] Ubit Umarov: not many bug reports
[11:56] Ubit Umarov: one i still can't repo
[11:56] Gavin.Hird @grid.xmir.org:8002: not really
[11:57] Andrew Hellershanks: Gavin, I do have a viewer question.
[11:57] Gavin.Hird @grid.xmir.org:8002: LL said they will add new features that requires viewer changes this year
[11:57] Gavin.Hird @grid.xmir.org:8002: no mention on what
[11:57] Gavin.Hird @grid.xmir.org:8002: sure
[11:57] Ubit Umarov: gavin that was "no news" :)
[11:57] Ubit Umarov: part of meeting
[11:57] Ubit Umarov: ofc they will add something
[11:58] Gavin.Hird @grid.xmir.org:8002: new features are stuff like Bento or soemthing
[11:58] Ubit Umarov: bet they have a ton of feature requests also
[11:58] Gavin.Hird @grid.xmir.org:8002: real viewer cahnges
[11:58] Andrew Hellershanks: Sometimes when I enter a coordinate for a prim in the Build dialog it goes in as I typed it. Other times it has some extra non-zero digit added. I can force it back to the three decimal places I entered by changing the number then changing it back. Any ideas why that happens?
[11:59] Ubit Umarov: Kb to chair interface issue?
[11:59] Andrew Hellershanks: The viewer is quite capable of handling 3 decimal places. Always a pain when it wants to add some extra digit or two.
[11:59] Gavin.Hird @grid.xmir.org:8002: most likely some floating point inaccuracy
[11:59] Andrew Hellershanks: Yes, but if I enter some other value then enter what I originally entered it usually sticks.
[11:59] Gavin.Hird @grid.xmir.org:8002: I have seen it occasionally in my viewer
[11:59] Andrew Hellershanks: yea. Doesn't happen all the time but it is annoying.
[11:59] Ubit Umarov: yeha some numbers back from float are not as we typed
[12:00] Ubit Umarov: bc have no exact representation
[12:00] Gavin.Hird @grid.xmir.org:8002: which viewer is this?
[12:01] Ubit Umarov: but a round bf display should hide that
[12:01] Andrew Hellershanks: Some won't but if I'm able to put in another number then what I wanted and it sticks that seems to indicate it isn't always a representation issue.
[12:01] Andrew Hellershanks: Gavin, um... I think all the ones I've used.
[12:01] Andrew Hellershanks: Most recently in Firestorm.
[12:01] Kayaker Magic: I've had a problem where the viewer just doesn;t display all the digits in a number, if you scroll the little window left, you will see them
[12:02] Kayaker Magic: So I sometimes get a combination of those invisible digits added to what I type.
[12:02] Gavin.Hird @grid.xmir.org:8002: some of the code that handles formatting is ancient, and even the xml for how the UI is laid out can be different from field to filed depending on when it was originally generated
[12:02] Andrew Hellershanks: Kayaker, that can happen as the window only show so many digits at a time.
[12:03] Andrew Hellershanks: hehe... it should store the numbers in BCD format. No problem. ;)
[12:03] Ubit Umarov: no idea when we will actuall add a new feature to opensim :(
[12:03] Andrew Hellershanks: Did LL say what new features they are planning to add?
[12:04] Ubit Umarov: no andrew  was just a "sure we will have something"
[12:04] Andrew Hellershanks: ok
[12:04] Gavin.Hird @grid.xmir.org:8002: the xml that lays out a field in the UI can be stacked many ways, and depending on what comes before what, it is getting bounced around different functions, so yeah
[12:04] Ubit Umarov: and ll adding a feature does not mean we will :)
[12:04] Gavin.Hird @grid.xmir.org:8002: IDK
[12:05] Andrew Hellershanks: Ubit, of course not.
[12:05] Gavin.Hird @grid.xmir.org:8002: PBR is a big request
[12:05] JayR.Cela @grid.3rdrockgrid.com:8002: PBR ?
[12:05] Andrew Hellershanks: Gavin, Pabst Blue Ribbon??
[12:05] Gavin.Hird @grid.xmir.org:8002: physics based rendering
[12:05] Kayaker Magic: LOL
[12:05] Ubit Umarov: wee seen no feedback about 0.9.2.0 release
[12:06] Ubit Umarov: in fact not even coments
[12:06] Gavin.Hird @grid.xmir.org:8002: https://en.wikipedia.org/wiki/Physically_based_rendering
[12:06] Andrew Hellershanks: Gavin, ok. That makes more sense. Mind you, the coders might prefer some of the other version of PBR.
[12:06] Ubit Umarov: guess many grids where already on it
[12:07] Andrew Hellershanks: Ubit, I wouldn't count on that. A lot had not even upgraded to 0.9 or had been slow to adopt it. I wouldn't expect a fast migration to 0.9.2
[12:07] Gavin.Hird @grid.xmir.org:8002: I also suspect they may launch an xbox version of the viewer
[12:07] Michael.Christopher @cc.opensimulator.org:8002: LL xbox viewer?!?
[12:07] Ubit Umarov: physical base rendering is called "Bob Ross"
[12:07] Gavin.Hird @grid.xmir.org:8002: yeah xbox
[12:07] Ubit Umarov: laso includes "happy little trees" etc
[12:07] Gavin.Hird @grid.xmir.org:8002: which will reguire PBR
[12:08] Andrew Hellershanks: Ubit, :)
[12:08] Gavin.Hird @grid.xmir.org:8002: but we shall see
[12:08] Andrew Hellershanks nod
[12:09] Andrew Hellershanks: Any final thoughts, comments, or questions before we wrap up todays gathering?
[12:09] JayR.Cela @grid.3rdrockgrid.com:8002: Im good :_)
[12:09] Ubit Umarov: whats for dinner?
[12:09] Gavin.Hird @grid.xmir.org:8002: that was 3 hours ago
[12:09] Andrew Hellershanks: Stir Fry
[12:09] Michael.Christopher @cc.opensimulator.org:8002: im good too lol
[12:09] Jamie.Jordan @grid.kitely.com:8002: yeah i'm hungry
[12:10] Kayaker Magic: leftover roast duck
[12:10] Ubit Umarov: Dann and you didn't call me gavin ?
[12:10] Gavin.Hird @grid.xmir.org:8002: hehe
[12:10] Andrew Hellershanks: ok, I'll let you head out in search of food if you haven't already eaten.
[12:10] Gavin.Hird @grid.xmir.org:8002: I have some lunch trays bought on Kitely
[12:10] Andrew Hellershanks: That will do it for today. Thank you all for coming. See you again next week.