User:Babbage Linden/Office Hours/2010 05 12
Jump to navigation
Jump to search
Transcript
[03:00] | Ardy Lay: | Hi Karel |
[03:00] | Ardy Lay: | Hi Latif |
[03:00] | Jonathan Yap: | Think positive--at least you are not in a sailboat :) |
[03:00] | Karel Linden: | hi all |
[03:01] | Ardy Lay: | Hi Johnny |
[03:01] | Morgaine Dinova: | Hi Karel |
[03:01] | Johnny Linden: | Hi |
[03:01] | Phantom Ninetails: | Greetings Karel, and everyone else I haven't greeted |
[03:01] | Latif Khalifa: | oh, linden invasion today :) |
[03:01] | Morgaine Dinova: | Hi Johnny |
[03:01] | Johnny Linden: | Hi |
[03:01] | Ardy Lay: | Hi Babbage |
[03:01] | Phantom Ninetails: | All sorts of Lindens, yet no babb.. Oh wait there he is |
[03:01] | Latif Khalifa: | hey Babbage |
[03:01] | Phantom Ninetails: | Greetings Babbage |
[03:01] | Morgaine Dinova: | And Babbage in a skirt, cool :-) |
[03:02] | Kaluura Boa: | =^_^= |
[03:02] | Techwolf Lupindo: | Hi all |
[03:02] | Babbage Linden: | morning all |
[03:02] | Morgaine Dinova: | Oh, a Chinese sort of uniform garment finally materialized :-) |
[03:02] | Morgaine Dinova: | Hiya Babbage :-) |
[03:03] | Babbage Linden: | we have a mass linden invasion today |
[03:03] | Ardy Lay: | Hi Gayathri |
[03:03] | Jonathan Yap: | Big announcement? :) |
[03:03] | Latif Khalifa: | how come kelly is with us at this hour? ;) |
[03:03] | Phantom Ninetails: | Insanity, they just keep coming |
[03:03] | Babbage Linden: | as you may have noticed |
[03:03] | Ardy Lay: | Yes, but, is is really an invasion? :-) |
[03:03] | Babbage Linden: | the whole of the pixiedust team are in brighton for 2 weeks |
[03:04] | Ardy Lay: | Hi Gisele |
[03:04] | Babbage Linden: | so invited them along |
[03:04] | Latif Khalifa: | ah, that explains it |
[03:04] | Ardy Lay: | Oh! Cool! |
[03:04] | Jonathan Yap: | Is this Linden invasion like cramming people into a VW bug? |
[03:04] | Phantom Ninetails: | Ah |
[03:04] | Morgaine Dinova: | That's cool :-) |
[03:04] | Babbage Linden: | some of you may not have met everyone yet, so kelly want to introduce yourself? |
[03:04] | Latif Khalifa: | poor people of brighton :P |
[03:04] | Phantom Ninetails: | Feels that way for me Jonathan, eating up ram and pushing things into swap like crazy |
[03:05] | Babbage Linden: | kelly? what do you do at linden? |
[03:05] | Latif Khalifa: | nap ;) |
[03:05] | Morgaine Dinova: | Don't see Gisele yet, but hi Gayathri :-0 |
[03:05] | NikoKito Aries: | hello good morning |
[03:05] | Jonathan Yap: | Gisele was diving |
[03:05] | Gisele Linden: | its fun |
[03:05] | Morgaine Dinova: | Haha |
[03:05] | Babbage Linden: | ok, kelly's away |
[03:05] | Gisele Linden: | few cats like water |
[03:05] | Techwolf Lupindo: | heeh |
[03:05] | Gisele Linden: | heh |
[03:05] | Techwolf Lupindo: | Big cats like water. |
[03:05] | Babbage Linden: | gayathri, want to introduce yourself? |
[03:06] | Gayathri Linden: | Hi there, Sorry, was having some trouble with my keyboard |
[03:06] | Gayathri Linden: | I'm Gayathri Linden, and I'm a Program Manager at Linden working with Blighty |
[03:06] | Techwolf Lupindo: | Go wired. Wireless keyboards don't work so well. :-) |
[03:07] | Morgaine Dinova: | You know that nobody calls us Blighty anymore, outside of Lindens ;-) |
[03:07] | Babbage Linden: | (PixieDust nee Blighty - we're international now) |
[03:07] | Gayathri Linden: | :) PixieDust - mostly anything Babbage's team needs, I'm there for them. |
[03:07] | Babbage Linden: | Gayathri also PMs for a number of the other teams in Core |
[03:07] | Babbage Linden: | and also rides motorcycles |
[03:08] | Morgaine Dinova: | Hehe |
[03:08] | Xugu Madison: | (Morning all) |
[03:08] | Babbage Linden: | Kelly, want to introduce yourself? What have you done at Linden? |
[03:08] | Kelly Linden: | Hi! |
[03:08] | NikoKito Aries: | good morning Xugu |
[03:08] | Latif Khalifa: | hey Kelly :) |
[03:08] | Morgaine Dinova: | Pity Ceawlin's not here, she'd immediately relate to motocycles :-) |
[03:08] | Kelly Linden: | I'm a developer for linden for the past 6 years. Worked on back end stuff mostly, scripting to scalability. |
[03:09] | Gisele Linden: | tell 'em bout your casino! |
[03:09] | Kelly Linden: | ssshhh |
[03:09] | Latif Khalifa: | lol |
[03:09] | Techwolf Lupindo | lol |
[03:09] | Kelly Linden: | :) |
[03:09] | Babbage Linden: | *coughs* secrets *cough* |
[03:09] | Gisele Linden: | heh |
[03:09] | Gisele Linden: | had to... |
[03:09] | Babbage Linden: | ok, apart from outing people, what do you do Gisele? |
[03:09] | Gisele Linden: | I am the Product Point for Pixies Dust --my main goal in life is to work with Babbage on a roadmap to ensure its in line with where we want to go strategically as a company, and as a platform offering to developers. |
[03:09] | Latif Khalifa: | i notice all Babbages Oauth videos had some sort of gambling device as an example object :P |
[03:10] | Xugu Madison: | ...Pixies Dust? |
[03:10] | Latif Khalifa: | must be stolen from kell'y casino ;) |
[03:10] | Babbage Linden: | Scripts are the magic pixie dust that bring second life to life |
[03:10] | Babbage Linden: | hence pixie dust :-D |
[03:11] | Babbage Linden: | also, it fits in with out dandruff based project naming scheme: see also firefly and snowglobe |
[03:11] | Gisele Linden: | btw -I didnt come up with that |
[03:11] | Phantom Ninetails: | lol |
[03:11] | Latif Khalifa: | hehe |
[03:11] | Babbage Linden: | ok, Johnny, want to introduce yourself? |
[03:12] | Xugu Madison: | Babbage, did you bring half the lab? :) |
[03:12] | Morgaine Dinova: | I hope "where Linden wants to go strategically as a company" matches where SL residents want to live :-) |
[03:12] | Johnny Linden: | Hi. I'm a Core dev working on scripting. Previusly part of WebCore working on something I'm not sure I can talk about. I like to embarras the impossible. |
[03:12] | Johnny Linden: | and it seems I cannot spell today. |
[03:12] | Babbage Linden: | :-D |
[03:12] | Johnny Linden: | all wolf. all the time. |
[03:12] | Gayathri Linden: | embarrass, or embrace? :) |
[03:12] | Techwolf Lupindo: | TrO{gO DsA[M,bN HyAqR4tDc TgrOo TgYPmE WeIjTyH P;AzWqS,. * |
[03:12] | Johnny Linden: | both |
[03:13] | Phantom Ninetails: | lol |
[03:13] | Babbage Linden: | ok, Karel you met last week, but want to introduce yourself for the new people? |
[03:13] | Techwolf Lupindo: | Yea! I"me not the only werewolf here. lol |
[03:13] | Karel Linden: | Johnny is recovering from a sofa related injury |
[03:13] | Latif Khalifa: | thosa can be nasty ;) |
[03:13] | Babbage Linden: | Xtreme sofaing |
[03:13] | Gisele Linden: | omg that sounded bad |
[03:13] | Gisele Linden: | lot o typing goin on |
[03:13] | Latif Khalifa: | especially when beer is involved ;) |
[03:13] | Gayathri Linden: | apparently far more dangerous than motorcycling or skating |
[03:13] | Karel Linden: | I'm back working on Pixie Dust |
[03:13] | Babbage Linden: | you should have seen the ramp he was launching it off |
[03:14] | Latif Khalifa: | lol |
[03:14] | Phantom Ninetails: | lol |
[03:14] | Babbage Linden: | ok, I think that's everyone |
[03:14] | Karel Linden: | I've been doing Mono 2 and TCMalloc |
[03:14] | Xugu Madison | is frankly concerned by you lot :) |
[03:14] | Johnny Linden: | giseles cold beer directive is helping with my injury. |
[03:14] | Babbage Linden: | so it's great to have the whole team in Brighton |
[03:14] | Gisele Linden: | ;0) Product is worth *something* |
[03:14] | Morgaine Dinova: | Oooh, tell us more about TCMalloc :-) |
[03:14] | Techwolf Lupindo: | I've tried to build the viwer with tcmallac. Got it built, but suffered horriable performance. |
[03:15] | Babbage Linden: | as I mentioned last week, we're taking the time to run a hackathon to build C# demos |
[03:15] | Babbage Linden: | thanks for all your ideas and suggestions last week |
[03:15] | Babbage Linden: | we've followed them all up and come up with a set of really exciting demos we're working on now |
[03:15] | Karel Linden | notes request for TCMalloc info |
[03:15] | Babbage Linden: | the first is a set of performance benchmarks |
[03:16] | Babbage Linden | also notes it and will come back to it |
[03:16] | Babbage Linden: | kelly, want to talk about where you've got with that |
[03:16] | NikoKito Aries: | bienvenida karessa |
[03:16] | NikoKito Aries: | (es->en) Welcome karess |
[03:17] | karessa Karas: | hola, hi all |
[03:17] | NikoKito Aries: | hay sillas libres |
[03:17] | NikoKito Aries: | (es->en) chairs are free |
[03:17] | karessa Karas: | rezz |
[03:17] | Latif Khalifa | imagines using standard c# math stuff would give some great gains |
[03:17] | JB Hancroft: | gm all :) |
[03:17] | Latif Khalifa: | and hashing/encryption |
[03:17] | Gisele Linden: | say more |
[03:18] | Kelly Linden: | rough estimate from the benchmarks babbage used in the mono conversion is maybe 5% ish improvement doing C# |
[03:18] | Gisele Linden: | several areas to use math libraries --where were you thinking of using? |
[03:18] | Kelly Linden: | The real win is when you can use system libraries (like Math) instead of calling ll functions |
[03:18] | Kelly Linden: | Then you get a couple orders of magnitude performance improvement. Give or take an order |
[03:18] | JB Hancroft: | The encryption example is a good one. |
[03:19] | Latif Khalifa: | Gisele, trying to do hashes/encryption with LSL is torture, I can imagine thing like generating oauth otp would be 1000 times faster |
[03:19] | Babbage Linden: | We're looking in to whether we can use the encryption libraries |
[03:19] | Babbage Linden: | as they're not part of the EMCA BCL |
[03:20] | Babbage Linden: | so we need to find out from the Mono team whether they're covered by the Microsoft promises |
[03:20] | Xugu Madison | would also imagine C# has less memory overhead from having to work around the language, and the scripts will be less brittle... |
[03:20] | Opensource Obscure: | nikokito : ) ...shut that bloody chair please |
[03:20] | Babbage Linden: | but that's a great suggestion Latif, we have LSL and C# versions of the OAuth library that we can use as benchmarks |
[03:20] | Xugu Madison: | Babbage; you may also want to check what the legal status of providing encryption tools is |
[03:20] | Gisele Linden: | yep -great point |
[03:21] | Babbage Linden: | and which will be more relevant than some of the abtract benchmarks we have |
[03:21] | Babbage Linden: | we've also been thinking about profiling in space |
[03:21] | Latif Khalifa: | and these days apis and conneting to all sort of services are all the hoot ;) |
[03:21] | Babbage Linden: | which will require some changes to the simulator to do proper high tide memory tracking |
[03:22] | Babbage Linden: | yes, Latif, that brings us neatly on to the demos that Johnny as been working on |
[03:22] | Babbage Linden: | want to talk about those Johnny? |
[03:22] | Johnny Linden: | Using the gdata api to poke and peek data in a google spreadsheet. |
[03:23] | Techwolf Lupindo: | Combat systems regually hit performace problems as they talk to http server at high speed. |
[03:23] | Babbage Linden: | the nice thing about that is that we can take the existing C# client library for GData and pull it in to SL |
[03:23] | Babbage Linden: | so, instead of having to build a new client library for complex web services |
[03:23] | Babbage Linden: | (like the OAuth library) |
[03:24] | Babbage Linden: | you already have a convinient client layer |
[03:24] | Latif Khalifa: | hmm, aren't you going to provide ll* version of http fetch? |
[03:24] | Johnny Linden: | You could have the spreadsheet do some wacky calculations then read the result back. not a great way to do complex calc offload, tho. |
[03:24] | Xugu Madison | oooooohs, and was playing with GData for work yesterday as it happens |
[03:24] | Babbage Linden: | Latif, we'd like to enable the .NET APIs for HTTP |
[03:24] | Babbage Linden: | as it makes porting these libraries much simpler |
[03:25] | Latif Khalifa: | oh, that would be very very nice :) |
[03:25] | Babbage Linden: | we're going to use .NET for the demos as it makes it easier for us |
[03:25] | Babbage Linden: | (and you can imagine porting the GData client lib to llHttpRequest if you had to) |
[03:25] | Latif Khalifa: | yeah, that's the first thing that popped into my head ;) |
[03:25] | Babbage Linden: | there's still a lot of thinking to be done there, but for now we're going to do the easiest thing that gets us some demos |
[03:26] | Latif Khalifa: | what c# language version are we going to get? |
[03:26] | Babbage Linden: | the final demo we're working on was another idea of Johnnys |
[03:26] | Babbage Linden: | which is to build a profiling tool for scripts |
[03:26] | Gayathri Linden: | love johnny for tying it together, btw. |
[03:26] | Morgaine Dinova: | You actually do mean .NET, as opposed to Mono? |
[03:26] | JB Hancroft: | profiling tool... I like that a lot :) |
[03:27] | Johnny Linden | is worried he'll get a rep as a big picture kind of guy. |
[03:27] | Latif Khalifa: | Morgaine, mono is an implementation of .NET |
[03:27] | Babbage Linden: | it uses the delegate dispatch mechanism we've built for C# event handling |
[03:27] | Babbage Linden: | to build C# decorator style extensions |
[03:27] | Babbage Linden: | you instantiate a Profiler object and it replaces the event handler delegates for the script |
[03:28] | Babbage Linden: | with alternative handlers which reset the time, call the original delegate and then check the time |
[03:28] | Morgaine Dinova: | Latif: .NET is MS's implementation. If you say "using .NET" when most of the discussion is about Mono, then that usually means you want to use MS code. I'm asking for clarification. |
[03:28] | JB Hancroft: | does that get driven by timers, then? or by looking at the time when dispatch events happen? |
[03:28] | Kelly Linden: | basically a drop in profiler that doesn't require modifying the existing handlers |
[03:29] | Babbage Linden: | what you get is the time it took to execute the handler |
[03:29] | Latif Khalifa: | are we going to get c# 3.0? (please) with things like llObject.Clicked += (object sender, ClickedEventArgs e) { llSay(0, "Hello " + e.DetectedName[0]); } |
[03:29] | Babbage Linden: | we're building another which logs event parameters |
[03:29] | Morgaine Dinova: | No, it's C# 1.0 Latif |
[03:29] | Babbage Linden: | and you can chain the 2 together to get logging and profiling if you want |
[03:29] | JB Hancroft: | for tracing...? |
[03:29] | Babbage Linden: | the nice thing is you can just drop either in to an existing script |
[03:30] | Babbage Linden: | so, you don't need to add manual timing and tracing code to each method |
[03:30] | Babbage Linden: | it feels a lot like aspect oriented programming |
[03:30] | JB Hancroft | nods... sounds very nice |
[03:30] | Babbage Linden: | and is very neat |
[03:30] | Latif Khalifa: | yeah |
[03:30] | Babbage Linden: | so, those are the planned demos which we're working on at the moment |
[03:31] | Babbage Linden: | we're also using the development time to collect bugs and limitations with the C# implementation and our API |
[03:31] | JB Hancroft: | was there an answer to which C# version we're getting? perhaps I missed that. |
[03:31] | Babbage Linden: | and to test drive the docs |
[03:31] | Gisele Linden: | C#1.0 |
[03:31] | Babbage Linden: | we're currently supporting C# 1.0 |
[03:31] | Latif Khalifa: | aww :( |
[03:31] | Morgaine Dinova: | Babbage: and you're going to use them with non-Mono libraries, for now at least? |
[03:31] | Kelly Linden: | It would also be easy to build a profiler that ran through and exercised all the handlers, used some reflection to exercisize other functions even and could isolate the performance bottlenecks etc. |
[03:32] | Babbage Linden: | as it allows us to ship without having to rewrite the UThreadInjector |
[03:32] | Babbage Linden: | hopefully we will get to C# 2.0 later |
[03:32] | Babbage Linden: | if gisele can convince people we should support it :-D |
[03:32] | Gisele Linden: | which depends on how compelling C#2.0 is |
[03:32] | Babbage Linden: | Morgaine, we're only using non-Mono libraries for the demos |
[03:32] | Latif Khalifa: | well making c# 1.0 only in 2010 is kind of... retro ;) |
[03:33] | Gisele Linden: | we had that discussion |
[03:33] | Morgaine Dinova: | Babbage: kk, tnx |
[03:33] | Babbage Linden: | the plan is to support ECMA-BCL first |
[03:33] | JB Hancroft: | retro has a certain appeal ;) |
[03:33] | Babbage Linden: | but, we want to make demos that show where we could go in the future |
[03:33] | Gisele Linden: | but there are good reasons for it in terms of our timeline and get C# functionality out there sooner than *never* |
[03:33] | Gisele Linden: | -right |
[03:33] | Karel Linden: | 1.0 in 2010 is better than 3.0 in 2011 |
[03:33] | Babbage Linden: | and we need to talk to people about the legalities of going outside the ECMA-BCL |
[03:34] | Babbage Linden: | also, C# 1.0 is a lot less retro than LSL |
[03:34] | Babbage Linden: | doing the kind of delegate interception and reflection we're doing in these demos is way beyong LSL |
[03:34] | Babbage Linden: | (which is kind of the point, as we're demoing the improvements C# gives us here) |
[03:34] | Latif Khalifa: | yes, but reusing code out there is going to be difficult with c# 1.0 only... |
[03:35] | Babbage Linden: | Latif, yes, that's what we're finding |
[03:35] | Babbage Linden: | so the investigations were useful |
[03:35] | Babbage Linden: | and give us a much stronger case for C# 2.0+ |
[03:36] | Babbage Linden: | so, that's where we are with C# |
[03:36] | JB Hancroft: | Would the effort in testing and supporting C# 1.0 be better spent getting to 2.0+ sooner? |
[03:36] | Babbage Linden: | hopefully we will have some neat demos to how you soon |
[03:36] | Babbage Linden: | JB, there's not a lot of wasted work |
[03:36] | Babbage Linden: | mostly what we need to support C# 2.0 is a ported UThreadInjector |
[03:37] | Babbage Linden: | and that porting won't be a lot more work having shipped C# 1.0 |
[03:37] | Latif Khalifa: | c# 1.0 doesn't even do List<string> generics, does it? |
[03:37] | Morgaine Dinova: | Does each major version of C# have a corresponding runtime? I know it doesn't matter initially since we'll be uploading C# source to you, but eventually when you accept CLI assemblies from us, it'll make a difference if the CLR has a C#-version specific runtime. |
[03:37] | Babbage Linden: | Morgaine, I think we'll want to accept .NET 2.0 assemblies before we allow uploading |
[03:37] | JB Hancroft: | Do you have some rough estimate on the amount of time, after 2.0, that it would take to get us to 3.0? |
[03:37] | Babbage Linden: | there are only really 2 versions of the assemblies, before generics and after |
[03:37] | Kelly Linden: | by the time we allow uploading of bytecode directly you won't be restricted to C#1.0 |
[03:38] | Babbage Linden: | JB, 3.0 doesn't add a lot to the runtime |
[03:38] | Babbage Linden: | but it's also not ECMA standardised... |
[03:38] | Morgaine Dinova: | Thanks, I get the general timeline then :-) |
[03:38] | Babbage Linden: | so, you should prod MS about that ;-) |
[03:38] | JB Hancroft: | heh... ol |
[03:38] | JB Hancroft: | *ok |
[03:38] | Latif Khalifa: | not having generics really kills standard collections |
[03:39] | Babbage Linden: | ArrayLists are still useful (we use them for LSL lists) |
[03:39] | Babbage Linden: | but I agree, generic collections are better |
[03:39] | Babbage Linden: | you still get typed arrays... ;-) |
[03:39] | Latif Khalifa: | yeah, good, we're gettin 2002 level tech :) |
[03:40] | Jonathan Yap: | Later on will you still have the staff to move to 2.0/3.0 or will they have gone on to other projects? |
[03:40] | Gisele Linden: | ouch |
[03:40] | Babbage Linden: | which is better than 2003 level game scripting tech it turns out ;-) |
[03:40] | Latif Khalifa: | it is |
[03:40] | Babbage Linden: | it really does make a big difference |
[03:40] | Latif Khalifa: | still disappointing |
[03:40] | Babbage Linden: | SL feels like a proper development platform |
[03:41] | Latif Khalifa: | knowing LL "we'll do it in the future" means 5 years at least ;) |
[03:41] | Babbage Linden: | well, hopefully Gisele can avoid that |
[03:41] | JB Hancroft | hands Latif a happy pill ;) |
[03:41] | Babbage Linden: | I would certainly like to |
[03:41] | Morgaine Dinova: | Babbage works faster than Nyx at least. Nyx's favourite date is 2027 ;-) |
[03:41] | Gisele Linden | wants to know where to get those happy pills |
[03:42] | Babbage Linden: | also, 2002 level .NET ~= 2010 Java |
[03:42] | Gisele Linden: | BURN |
[03:42] | JB Hancroft: | We can hook you up ;) |
[03:42] | Babbage Linden: | so you could look at it that way ;-) |
[03:42] | Latif Khalifa: | java is c# 0.5 ;) |
[03:42] | Babbage Linden: | ok, any more C# thoughts? |
[03:42] | Morgaine Dinova: | Yeah. Debugging. |
[03:42] | Techwolf Lupindo: | LUA |
[03:43] | Morgaine Dinova: | How are we gonna do it, once we have C# scripting? |
[03:43] | Johnny Linden | <3 Lua |
[03:43] | Babbage Linden: | Debugging is going to be hard inside the simulator |
[03:43] | Latif Khalifa | goes to sob quetly in the corner and dream of List<UUID> :P |
[03:43] | Gayathri Linden | hands latif a box of tissues. :) |
[03:43] | Babbage Linden: | But, having C# is going to make developing in IDEs much easier |
[03:43] | Morgaine Dinova: | <--- <3 Lua too. Lua should have been used for LSL-NG, not C# ;-) |
[03:44] | Babbage Linden: | I would like to see a fully mocked set of LindenLab.SecondLife libraries that people can use for unit testing and debugging |
[03:44] | Ardy Lay | tries to ignore the weather warning sirens. (RL) |
[03:44] | Babbage Linden: | running untrusted code inside your process is scary enough without letting untrusted people run debuggers within your process |
[03:44] | Babbage Linden: | and lua seems nice |
[03:45] | Latif Khalifa: | google app engine provides a set of mock libraries for local coding/debugging |
[03:45] | Babbage Linden: | is there an official GData client library for it? |
[03:45] | Babbage Linden: | Latif, right, I would like to do that |
[03:45] | Babbage Linden: | and I'm encouraged that they've taken the same tack |
[03:46] | Babbage Linden: | (gisele, put it in our reasons to open source the API code doc) |
[03:46] | Johnny Linden: | a simulator simulator |
[03:46] | JB Hancroft: | Babbage - is there any particular IDE that we'll be favoring, over another? |
[03:46] | Latif Khalifa: | perhaps we can get LindenLab.SecondLife.dll that can do at least do all the syntax |
[03:46] | Phantom Ninetails: | Last I checked their servers don't run Windows.. |
[03:47] | Babbage Linden: | JB, I'm sort of hoping that we can open source our APIs (or build the mocks) and have IDE developers rush to add SL support ;-) |
[03:47] | Morgaine Dinova: | I wish some new debugging system were added to SL scripting at the same time as the move to C#. Some nice way of externalizing the state of scripts. It's not like running C# locally, gonna be tough without server-side support. |
[03:47] | Babbage Linden: | MS were actually keen a while ago, but that was in the middle of the hype spike... |
[03:47] | JB Hancroft: | ok |
[03:47] | JB Hancroft: | probably something the Emerald team will step up to. |
[03:48] | Babbage Linden: | Morgaine, we're currently working out whether we can safely allow access to Serialization |
[03:48] | Babbage Linden: | (Silverlight doesn't allow it) |
[03:48] | Babbage Linden: | if we do, you can just serialize your script and HttpRequest it somewhere else... |
[03:48] | Babbage Linden: | (which would also be a neat thing to do in a Profiling style decorator...) |
[03:49] | Morgaine Dinova: | Not sure I get how that would work for debugging |
[03:49] | JB Hancroft: | a small wrinkle for script permissions... |
[03:49] | Latif Khalifa: | having a public api to leverage existing IDEs would be awesome |
[03:49] | Babbage Linden: | Latif, yes, that could either be a client or server API |
[03:50] | Babbage Linden: | having a viewer launch a decent IDE would be nice |
[03:50] | Latif Khalifa: | I mean having an assembly that supports ll* API |
[03:50] | Babbage Linden: | yes, I agree |
[03:50] | Babbage Linden: | a set of mocks that record calls, can have expectations set etc. |
[03:50] | Babbage Linden: | allow you to poke events at your script and make sure the right things happen |
[03:51] | Babbage Linden: | debugging sucks, testing rocks, as someone else may have said before |
[03:51] | Latif Khalifa: | :) |
[03:51] | Babbage Linden: | ok, we have 10 minutes left, want to give an update on TC Malloc Karel? |
[03:52] | Gayathri Linden: | w00t, yay tcmalloc! :) |
[03:52] | Karel Linden: | the original driver for TCMalloc was to enable profiling |
[03:52] | Karel Linden: | so if a region bloats we can find out why |
[03:52] | Karel Linden: | it may also be faster |
[03:53] | Karel Linden: | I've done dev tests and it seems to work |
[03:53] | Babbage Linden: | are you working with oskar on the beta on aditi? |
[03:53] | Karel Linden: | We're beta testing on a variety of hardware to check that still holds. |
[03:54] | Latif Khalifa | would love to peek into LL code that does assembly loading in app domains, it's soo easy to lock it and make it unloadable ;) |
[03:54] | Latif Khalifa: | that's 1.42 server? (tcmalloc) |
[03:54] | Karel Linden: | babbage: Yes |
[03:55] | Gayathri Linden: | babbage: yes (re oskar, beta) |
[03:55] | Gayathri Linden: | /latif - its slated for 1.42, yes. |
[03:55] | Morgaine Dinova: | Is the documentation for C# scripting still projected for next quarter? |
[03:55] | Gisele Linden: | this quarter, actually |
[03:55] | Babbage Linden: | gayathri and karel are working on shipping TC malloc in 1.42 |
[03:55] | Gisele Linden: | documentation... |
[03:56] | Gayathri Linden: | /gisele, end of this quarter, right? for documentation? |
[03:56] | Morgaine Dinova: | Woot, coming out before summer? Neat! |
[03:56] | Gisele Linden: | we are hoping to get our initial draft out by June-ish |
[03:56] | Gisele Linden: | and looking forward to feedback! |
[03:56] | Morgaine Dinova: | Very cool! |
[03:57] | JB Hancroft: | excellent :) |
[03:57] | Gisele Linden: | this will directly feed into our actual C# beta slated for the fall |
[03:57] | JB Hancroft: | things seem to be speeding up in terms of things coming out... hope the sr. mgr/handlers are being brutal |
[03:57] | Morgaine Dinova: | Will be nice to get back into SL scripting. LSL really put a big damper on it :-( |
[03:58] | Phantom Ninetails: | When do you think the version of debian used on the servers will be updated? |
[03:58] | Morgaine Dinova: | Fall == autumn, right? |
[03:59] | Babbage Linden: | yes |
[03:59] | Techwolf Lupindo: | While tcmalloc will enable lots of stuff for the Lindens, any benifent for end users? |
[03:59] | Morgaine Dinova: | Ta |
[03:59] | JB Hancroft: | *eep! .... "are not being brutal" :) |
[03:59] | Babbage Linden: | techwolf it will feed in to performace work |
[03:59] | Latif Khalifa: | i'm hoping we can get to mono 2.6.x soon, to kill the sim freezer |
[03:59] | Babbage Linden: | we'll use it to identify the causes of memory bloat |
[03:59] | Babbage Linden: | fix them and reduce lag |
[04:00] | Babbage Linden: | which will be the benefit most people are looking for :-D |
[04:00] | Babbage Linden: | ok, time's up |
[04:00] | Latif Khalifa: | yes |
[04:00] | Babbage Linden: | thanks for coming everyone! |
[04:00] | Phantom Ninetails: | Dang, no answer |
[04:00] | Techwolf Lupindo: | I want Linden bears!! |
[04:00] | Gisele Linden: | good chatting |
[04:00] | JB Hancroft: | thanks Babbage |
[04:00] | Techwolf Lupindo: | :-D |
[04:00] | Babbage Linden: | hope to see you all next week |
[04:00] | Ardy Lay: | Would be funny if the bloat never appears when using tcmalloc |
[04:00] | Xugu Madison: | Thanks for hosting Lindens! |
[04:00] | Gisele Linden: | later gaters |
[04:00] | Latif Khalifa: | nice talking to you guys too :) |
[04:00] | JB Hancroft: | is Kelly still here? |
[04:00] | Babbage Linden: | (talk to green re: debian) |
[04:00] | Xugu Madison | too would love any Linden bears.... |
[04:00] | Kelly Linden: | I am JB |
[04:00] | Morgaine Dinova: | Cyu Lindens |
[04:00] | Phantom Ninetails: | green? |
[04:00] | JB Hancroft: | ah, cool |
[04:00] | Kelly Linden: | sup? |
[04:01] | JB Hancroft: | I had a quick question |
[04:01] | JB Hancroft: | when we call llSetPrimitiveParamss |
[04:01] | JB Hancroft: | do updates to an object position get pushed all the way back ... |
[04:01] | Gayathri Linden: | Phantom: green linden |
[04:01] | JB Hancroft: | further than the sim the object is in? |
[04:01] | JB Hancroft: | it doesn't go back to the asset server, does it? |
[04:02] | Kelly Linden: | oh |
[04:02] | Kelly Linden: | no |
[04:02] | JB Hancroft: | what about changes in scale, etc? |
[04:02] | Kelly Linden: | I have a blurb on the wiki somewhere about the split between in world, inventory and asset |
[04:02] | Phantom Ninetails: | does green linden have an office hour or something |
[04:02] | JB Hancroft: | wiki: ok, cool. I'll go hunt for it. thanks. |
[04:03] | Kelly Linden: | nothing that is currently simulated (is in world) is saved in real time anywhere - it exists in simulator memory |
[04:03] | Kelly Linden: | once an hour the simulator saves it's entire state (including the state of every script and where every object is) to a file that is kept with the assets |
[04:03] | JB Hancroft: | ok, good... I was hoping it wasn't making the long trip |
[04:03] | Kelly Linden: | that state includes not jsut where objects are but every detail about the object |
[04:04] | Kelly Linden: | so at run time those LSL calls don't modify anything besides what is in simulator memory |
[04:04] | JB Hancroft: | ok :)) |
[04:05] | Morgaine Dinova: | I don't suppose there's a llFlush() function to affirm that current object state is recorded in case of hiccup. |
[04:06] | Kelly Linden: | There is not, we would have difficulty maintaining such a system |
[04:06] | Kelly Linden: | It could easily cause significant load. |
[04:06] | JB Hancroft: | I was wondering if there was an opportunity to do something "lighter"... to indicate an object change which is being done for animation, and we expect it will change again soon... |
[04:06] | Kelly Linden: | Especially by novice but paranoid scripters |
[04:06] | Morgaine Dinova: | All vendors are paranoid. |
[04:06] | Kelly Linden: | yea. |
[04:07] | Latif Khalifa: | there is no problem to solve here, thingie with sim states works nicely |
[04:07] | JB Hancroft: | Of course Morgaine... they let people like me on the grid ;) |
[04:07] | Techwolf Lupindo: | Having a vender hickuping on a $20USD sale for every few transactions.....I think you can get the picture. |
[04:07] | Jonathan Yap: | Well, except for the case where a nocopy object is rezzed and the sim crashes soon afterwards; the object is lost |
[04:08] | Phantom Ninetails: | I always discourage selling with 'no copy' permissions. |
[04:08] | Phantom Ninetails: | Only serves to cause content loss. |
[04:09] | Kelly Linden: | actually most crashes save state anyway on crash. (some don't, but trying to save and upload a file in a signal handler is a long shot anyway) |
[04:09] | Morgaine Dinova: | "No copy" is a bit of a bizarre concept in VWs anyway, which have copying at tons of points in the system. And with interop, that'll be even more so." |
[04:09] | Kelly Linden: | I've gotta run. cya |
[04:09] | Jonathan Yap: | Thank you Kelly |
[04:09] | JB Hancroft: | bye Kelly. thanks |
[04:09] | Morgaine Dinova: | Cya Kelly |
[04:09] | Phantom Ninetails: | Fare well |
[04:09] | Latif Khalifa: | have fun Kelly :) |
[04:10] | Gayathri Linden: | later, all, thanks again for the welcome! :) |
[04:10] | Morgaine Dinova: | TC Gayathri :-) |
[04:10] | JB Hancroft: | Bye Gayathri ... take care |
[04:10] | Xugu Madison: | Eek, still logged in. Bye all! |
[04:10] | JB Hancroft: | Bye all... time to head home |
[04:10] | Latif Khalifa: | take care you all |
[04:10] | Morgaine Dinova: | Cyu Xu |
[04:10] | Morgaine Dinova: | Me too. Cya all |
Generated with SLog Wikifier