User:Babbage Linden/Office Hours/2009 02 11

Transcript of Babbage Linden's office hours:

[8:33]	Babbage Linden:	hi everyone
[8:33]	Babbage Linden:	sorry i'm late
[8:33]	Cale Flanagan:	and whos xml is that?
[8:33]	Arawn Spitteler:	Hi, Babbage.
[8:33]	Siann Beck:	Hi Babbage!
[8:33]	Babbage Linden:	unfortunately i've got to rush off to another meeting
[8:33]	WolfPup Lowenhar:	hey babbage
[8:33]	Nock Forager:	lol
[8:33]	Fake Fitzgerald:	hi Babbage
[8:33]	Babbage Linden:	but si linden from brighton is going to come along to chat for a while
[8:34]	Morgaine Dinova:	Cool!
[8:34]	Babbage Linden:	he works with me in brighton, so can talk to you about what we've been doing with script metrics
[8:34]	Siann Beck:	Cool
[8:34]	Cale Flanagan:	do he has a bear?:)
[8:34]	Babbage Linden:	everything should be back to normal next week :-D
[8:35]	Arawn Spitteler recalls the meaning of SNAFU
[8:35]	Siann Beck:	Hey, Si!
[8:35]	Arawn Spitteler:	Hi, Si
[8:35]	Si Linden:	Hi Everyone
[8:35]	Nock Forager:	Hi Si.
[8:35]	Babbage Linden:	ok, have fun everyone
[8:35]	Fake Fitzgerald:	hi Si
[8:35]	Siann Beck:	Bye Babbage!
[8:35]	Cale Flanagan:	si, hi
[8:35]	Morgaine Dinova:	Still got snow in Brighton? It refuses to go away, just north of London
[8:35]	Babbage Linden:	i'll see you next week
[8:35]	WolfPup Lowenhar:	morning si
[8:35]	Babbage Linden:	code sage
[8:35]	Babbage Linden:	safe
[8:35]	Si Linden:	Wow - you still have snow? Ours has melted now.
[8:35]	Arawn Spitteler:	Safe is Sage?
[8:35]	Si Linden:	It was great while it lasted though
[8:36]	Morgaine Dinova:	Yeah, liked it. Beats rain every day
[8:36]	Si Linden:	so how are you all doing?
[8:36]	Morgaine Dinova:	Great Si -- how goes?
[8:36]	Siann Beck:	Not bad!
[8:36]	Arawn Spitteler's trying to recruit a few million dollars, but will surf the web a while more.
[8:36]	Si Linden:	glad to hear it :)
[8:37]	Si Linden:	hehe - good luck with that Arawn!
[8:37]	Morgaine Dinova:	Doing anything fun, Si?
[8:37]	Si Linden:	I am working on some performance testing at the moment
[8:37]	Morgaine Dinova:	Cool!
[8:37]	Si Linden:	it is good fun - and there are lots of nice graphs to look at :)
[8:37]	Morgaine Dinova:	What area?
[8:37]	WolfPup Lowenhar:	< has almost forgottne what snow looks and feels like :(
[8:37]	Morgaine Dinova:	Hehehe
[8:37]	Si Linden:	We have written a new internal web service that the sims will use
[8:38]	Si Linden:	and we need to know what sort of throughput it will handle gracefully
[8:38]	Si Linden:	So I am hammering it with requests!
[8:38]	Morgaine Dinova:	MRTG still lives? ;-)
[8:38]	Susie Chaffe:	gracefully as in not crashing the sim
[8:38]	Si Linden:	exactly :) sim crash == bad
[8:38]	Si Linden:	it should be totally isolated from the sim though
[8:38]	Morgaine Dinova:	I'm not voting for sim crash.
[8:38]	Arawn Spitteler:	Sim_Crash & Bad
[8:39]	Si Linden:	it will run on separate machines from any sim
[8:39]	Si Linden:	sim crashes are a big focus around here at the moment
[8:39]	Si Linden:	if I introduced some there would be big trouble :)
[8:39]	Morgaine Dinova:	Hehe
[8:39]	Siann Beck:	:)
[8:40]	Susie Chaffe:	any fallout from the Rerouted regions this morning
[8:40]	Morgaine Dinova:	Can't live in total safety and still progress though.
[8:40]	Si Linden:	It's going smoothly so far
[8:40]	Si Linden:	Morgaine: it is certainly a challenge, but we are really working hard to learn ways to make it work
[8:41]	WolfPup Lowenhar:	si is there a was to test netowrked vendors when the server is on agni and your on aditi?
[8:41]	Susie Chaffe:	something has been playing hookey with inworld emails for the last hour
[8:41]	Si Linden:	WolfPup: none that I know of, I believe the grids are totally separated
[8:41]	Arawn Spitteler:	Intergrid e-mail and http should be functional, in principal
[8:41]	Si Linden:	they are backed up by their own databases and routing tools
[8:42]	WolfPup Lowenhar:	ok
[8:42]	Si Linden:	but if you use external comms - like http you might be able to manage it
[8:42]	Si Linden:	Arawn: have you managed to get that workgin?
[8:42]	Morgaine Dinova:	I haven't looked at the instrumentation side of the client. When the Stats Bar says a Ping Sim time, is that a conventional ICMP echo, or something else?
[8:43]	Siann Beck:	Are there sims on agni with http-in?
[8:43]	Arawn Spitteler:	I don't know what an http_in might do, so I haven't tested it yet.
[8:43]	Si Linden:	http-in is not on agni yet, but it is going through the final stages of releasing
[8:43]	Si Linden:	so it should be real soon now :)
[8:43]	Arawn Spitteler:	As far as I know, only on Aditi. The software might not have passed Asset QA
[8:43]	Siann Beck:	Great!
[8:43]	Morgaine Dinova:	That's cool
[8:43]	Si Linden:	Morgaine: I am not sure, possibly, I will check
[8:43]	WolfPup Lowenhar:	i was wanting to work on moding the fossl vendor system on aditi so that i didn't have to wory about hving them cause problems on agni
[8:44]	Si Linden:	Morgaine: in fact I suspect it uses our own messaging mechanism
[8:44]	Morgaine Dinova:	Would make sense
[8:44]	Si Linden:	since it pings the simulator, not the host the sim is running on
[8:44]	Si Linden:	and we have some code in there already to check the aliveness
[8:45]	Si Linden:	so I guess we use those stats
[8:45]	Si Linden:	WolfPup: that sounds really sensible, I will do some research into inter-sim comms and ask babbage to let you know next week
[8:46]	Si Linden:	sorry - I mean inter-grid, not inter-sim
[8:46]	Imaze Rhiano:	so are scripts metrics coming on? can you yet estimate how much memory we will get for parcel / avatars?
[8:46]	Si Linden:	we have got script metric gathering code live on the agni grid now
[8:46]	Si Linden:	we are in the process of gathering the data
[8:46]	Si Linden:	ready for a proper analysis
[8:47]	Si Linden:	I believe that the intial figures are showing a 10% figure for mono
[8:47]	Si Linden:	i.e. 10% of all the scripts running anywhere are already Mono
[8:47]	Si Linden:	which is a really exciting figure and shows a good adoption rate
[8:47]	Morgaine Dinova:	Does the sound data (excluding Music which doesn't come from LL sim) take up much bandwidth? (I assume it's low samplerate for ambient and sounds)
[8:48]	WolfPup Lowenhar:	any scripts i write are mono and i recomplied the fossl vending system to mono
[8:48]	Si Linden:	Morgaine: I am not sure of the specific ratios of bandwidth, I know we have function in this area, but not sure what - I'll have a dig in our internal docs
[8:49]	Si Linden:	Gold star to you WolfPup :) that's great news
[8:49]	Morgaine Dinova:	Don't worry, though you might see it from your graphs
[8:49]	Nock Forager:	Is there any plan to release the detail you examined? Just curious, I want to know what type of scripts used much in this world.
[8:49]	Cale Flanagan:	poseballscripts;)
[8:50]	Nock Forager:	:)
[8:50]	Imaze Rhiano:	bedposeballscripts
[8:50]	Cale Flanagan:	y, i dodnt want to be to detailed:)
[8:50]	Nock Forager:	hehe just curious :)
[8:50]	Imaze Rhiano:	hairresizing scripts - in all those 255 prims
[8:50]	Arawn Spitteler heard something, about inactive scripts taking up time slices; if my polymorphic teleporters were functional, they use lists of scripts to restart.
[8:50]	Si Linden:	Hmm, re bandwidth ratios: I don't see anything on our wiki, I'll research and ask babbage to talk about it next week
[8:50]	WolfPup Lowenhar:	if the items useing those script are modable and the scripts are modable you can recomple them to mono easily
[8:51]	Siann Beck:	This might be a bit out of your jurisdiction, but any idea when we'll get Mono compile from inventory?
[8:51]	Si Linden:	re publicising script data: I am sure that we will make them public once we have verified them etc
[8:51]	Arawn Spitteler imagines recompiling scripts in each hair prim, to mono...
[8:52]	Nock Forager:	nice, I love metrics :)
[8:52]	Cale Flanagan:	easy, there is a mass recompile function
[8:52]	Si Linden:	me too Nock :) metrics + science rules!
[8:52]	Arawn Spitteler:	Mono takes a long time to recompile a script
[8:52]	WolfPup Lowenhar:	arwin all you have to do is edit the item and if there is a script in the root then you can recomple the while thing with no problem
[8:52]	Cale Flanagan:	(but do it works in all prims...)
[8:53]	Si Linden:	That's true Arawn, I have noticed it takes a little longer too
[8:53]	Morgaine Dinova:	Is anyone from Brighton going to be involved with the MMOX workgroup at the IETF? We're having quite a bit of fun on the mailing list already.
[8:53]	Si Linden:	Morgaine - I'll ask, but I haven't heard of anyone that is going yet
[8:54]	Si Linden:	Siann: mono compile from inventory - you mean so you don't have to rez stuff to recompile?
[8:55]	Siann Beck:	Right.
[8:55]	Si Linden:	I don't know of a plan for that, but I will ask babbage and get him to talk about it next week
[8:55]	Siann Beck:	Well, no, I mean so you can compile a script item in inventory.
[8:55]	Si Linden:	(I am building a little list for him eh?!)
[8:55]	Siann Beck:	:)
[8:55]	WolfPup Lowenhar:	i think when we select new script in the invetory can it be madt to slect mono
[8:55]	Cale Flanagan:	and put it back in inv to put it back into object to get same asset (code sharing)
[8:55]	Imaze Rhiano:	ya
[8:55]	Arawn Spitteler:	Inventory doesn't exist in sim, where the compilation is done, but all scripts are in inventory of something
[8:56]	Imaze Rhiano:	remember to mention that he will get 2 angry kitten to next meeting if he doesn't answer to them :P
[8:56]	Morgaine Dinova:	Hahaha
[8:56]	Si Linden:	hehe
[8:56]	Si Linden:	I will definitely tell him that!
[8:56]	Cale Flanagan:	they put already scratches in his stuff...
[8:57]	Si Linden:	Arawn: nice description, thanks - couldn't have said it better!
[8:57]	Morgaine Dinova:	Do you use a direct LSL to CLR compile on the sim, or translate to C# and then compile?
[8:58]	Arawn Spitteler doesn't think we've a C# compiler, yet
[8:58]	Si Linden:	We do a direct compile to mono byte code
[8:58]	Morgaine Dinova:	kk
[8:58]	Si Linden:	from the lsl text
[8:58]	Arawn Spitteler:	It's done bby the Simulator, which is why it takes longer, but how is this going to fly on Open Grid?
[8:58]	Imaze Rhiano:	Babbage told before that you have prototype C# compiler too
[8:59]	Imaze Rhiano:	one feature request: http://jira.secondlife.com/browse/SVC-2998
[8:59]	Darwin Recreant:	are we ever going to get client side syntax check back? compiling takes awhile just to check syntax
[8:59]	Morgaine Dinova:	That's actually why I asked the question Arawn --- I was going to follow up with "Is the translater going to be open sourced for compatibility on |interop?"
[8:59]	Morgaine Dinova:	interop*
[9:00]	Imaze Rhiano:	llWarpPos - warps object inside SIM to any position
[9:00]	Morgaine Dinova:	Because if there's no compatibility then interop kinda breaks.
[9:01]	Arawn Spitteler:	Anyother way to ask is, will InterOp create a market for the old LSL Bytecode?
[9:01]	Si Linden:	Fair point, I am not aware of a plan to do syntax verification in the viewer
[9:01]	Arawn Spitteler:	That might even be the question
[9:01]	Si Linden:	though I can see it would be really useful
[9:02]	Siann Beck:	Indeed it would.
[9:02]	Morgaine Dinova:	Last time we asked about syntax checking, the reply was "That's an ideal thing for the community to add to the viewer." And I think that;s right too.
[9:02]	Arawn Spitteler:	Syntax Verification is needed because Server Side Compilation is too slow. We have color flashes, viewer side, and these could be expannded.
[9:02]	Siann Beck:	There would still have to be support from LL's side though, Morgaine.
[9:02]	Si Linden:	Is there a JIRA for client side syntax verification?
[9:03]	Morgaine Dinova:	Siann: ideally they should supply the grammar parser at least, yeah.
[9:03]	Siann Beck:	Right.
[9:03]	Morgaine Dinova:	Is it done in yacc/bison?
[9:03]	Si Linden:	I know of no plan to open source our translator
[9:03]	Si Linden:	but that doesn't mean we don't have one :)
[9:03]	Imaze Rhiano:	http://jira.secondlife.com/browse/SVC-1401
[9:03]	Morgaine Dinova:	Hehe
[9:04]	Si Linden:	Morgaine: we certainly use lex and yacc for the lsl -> lsl bytecode
[9:04]	Si Linden:	but we use something a bit more flexible for lsl -> mono bytecode
[9:04]	Morgaine Dinova:	kk
[9:05]	Si Linden:	I can't think of the name of it right now
[9:05]	Morgaine Dinova:	If the lexer/parser is nice and separate, you could release just that, and let community build on it for syntax checking. Or, the old LSL compiler could be used of course.
[9:06]	Si Linden:	thanks Imaze - I suggest you all add your comments / requests to that JIRA to add weight to it!
[9:07]	Si Linden:	the lexer is the same for lsl + mono bytecode I think
[9:08]	Si Linden:	but the parser is modular
[9:08]	Si Linden:	and specific to each language (again *I think*!)
[9:09]	Si Linden:	I'll add that to babbages list - he has good links with the open sim crowd and probably has more info on this situation than I can guess :)
[9:09]	Morgaine Dinova:	Well here's the usual question then ... when are we getting O(1) arrays? ;-)))))))
[9:10]	Siann Beck:	Yes!
[9:10]	Si Linden:	lol Morgaine!
[9:10]	WolfPup Lowenhar:	most defantly
[9:10]	Si Linden:	so ... what is the usual answer?
[9:10]	Siann Beck:	LSL3?
[9:10]	Cale Flanagan:	usualanswer is, next server version:)
[9:11]	Imaze Rhiano:	when sky falls and elephants fly
[9:11]	Morgaine Dinova:	Well the answer from Don Linden 3 years ago was "Yeah, they're working fine, I just have a couple of corners to tweak." ;-))))
[9:11]	Siann Beck:	lol
[9:11]	Arawn Spitteler:	It's been promised in LSL#, as the reason for C#
[9:11]	Imaze Rhiano:	but you have C# prototype somewhere there? that should should solve problem...
[9:12]	Si Linden:	hehe - I don't know of active dev work in that area right now, we want to make sure Mono is fully stabilised before any major upheaval on the lsl lang
[9:12]	Si Linden:	We do have a c# prototype, it would be really gread to get that live
[9:12]	Arawn Spitteler:	Andrew's been waiting to add some features.
[9:12]	Si Linden:	it isn't fully functional, but works as a proof of concept
[9:13]	Morgaine Dinova:	I can't for the life of me understand why it's not been done, as it's such a small job, and would save lots of CPU cycles from not having to do all the silly list traversals.
[9:13]	Arawn Spitteler:	There should be some progressife documentation of C# features.
[9:13]	Si Linden:	Morgaine: I hear what you are saying, I can only assume that there is some difficulty with it
[9:13]	Si Linden:	maybe a 'quirck' in our language design
[9:13]	Arawn Spitteler:	PRIM_TYPE_TORTURED and llDetectedTarget are needed.
[9:14]	Si Linden:	otherwise I would guess it would have been done by now
[9:14]	WolfPup Lowenhar:	one thing is for shure arays would make the database script in the fossl vending system more functional
[9:15]	Imaze Rhiano:	Metadata verifier is coming in next Mono release (March) - so then you should drop it to that C# prototype and release it...
[9:16]	Si Linden:	That does sound sensible Imaze, but I can't make any guarantees of course
[9:16]	Arawn Spitteler:	C# on Aditi?
[9:17]	Cale Flanagan:	i would take C also, if c# not possible....
[9:17]	Si Linden:	hehe - there won't be a C version I am afraid!
[9:17]	Cale Flanagan:	(i will be the master of my 64k)
[9:17]	Arawn Spitteler:	C is too powerful; Scripting for Linden Servers has to be crippled
[9:18]	Imaze Rhiano:	how about Assembler? MC68000 or I486?
[9:18]	Imaze Rhiano:	:P
[9:18]	Siann Beck:	lol Imaze
[9:18]	Cale Flanagan:	y, mc68 would be acceptable
[9:18]	Arawn Spitteler wants a scriptable client
[9:18]	Darwin Recreant:	and plugins!
[9:18]	Cale Flanagan:	or c# and , how ist called so i can use pointers...
[9:18]	Si Linden:	We are waiting for the verifier to be avaiable in mono
[9:19]	Si Linden:	then we are going to porperly assess the c# scripting
[9:19]	Si Linden:	though, it would be significant work, and not likely to be shipped in the short term
[9:19]	Imaze Rhiano:	when are script memory limit going to hit?
[9:19]	Cale Flanagan:	just some corners...:P
[9:20]	Si Linden:	hehe - we could implement all sorts of assembly languages! That could be really fun!
[9:20]	Si Linden:	Imaze: we are gathering data right now, we need that data to make the decisions
[9:20]	Morgaine Dinova:	How is the wierd state machine system of LSL going to be expressed in C#?
[9:20]	Arawn Spitteler:	What is the general philosophic approach to InterOp ByteCode?
[9:21]	Si Linden:	Morgaine: there is certainly an element of hammering a square peg into a round hole!
[9:21]	Cale Flanagan:	i wouldnt cry,if i have no states in c#
[9:21]	Si Linden:	Arawn: as far as I know, bytecode is fully managed on the servers
[9:21]	Imaze Rhiano:	I think that Open SIM is using C# for scripting - likely similar model as there
[9:21]	Si Linden:	so I am not sure there is compatibility at that level at all
[9:22]	Si Linden:	only at the sourcecode level
[9:22]	Arawn Spitteler:	Last I attended Zero Hour, the talk was about protectig creativee authority, but this sounds like a question of the Flying Horses of Troy
[9:22]	Morgaine Dinova:	Si: what I really meant to ask was, is there going to be an outer-level state machine structure like in LSL, and just use C# for the functions, or is it going to be made to look like real C# files, and use special calls for the state machine functionality?
[9:22]	Morgaine Dinova:	Kind of opposite approaches
[9:23]	Si Linden:	ah i see Morgaine, I think that level of decision is still up for grabs
[9:23]	Morgaine Dinova:	kk
[9:23]	Cale Flanagan:	maybe u have to make ur own state machine, but who cares:)
[9:24]	Si Linden:	there is some info about the Mono Sandboxing tech that we are hoping will enable c# scripting: http://www.mono-project.com/MonoSandbox
[9:24]	Morgaine Dinova:	LSL is pretty unique in having an outer level that is a nicely readable state machine. (Pity the rest is so dire though, hehe :P)
[9:24]	Cale Flanagan:	if i could have state-local vars i would likethe states more...
[9:25]	Siann Beck:	That would definitely be nice, Cale.
[9:25]	Siann Beck:	I was just wishing for that on my latest project.
[9:25]	Cale Flanagan:	bec i really hate the mess with globals everywhere
[9:26]	Imaze Rhiano:	http://opensimulator.org/wiki/Scripting_Languages
[9:26]	Si Linden:	agreed Cale
[9:26]	Cale Flanagan:	(and there is already a jira as far as i remember)
[9:26]	Imaze Rhiano:	http://justincc.wordpress.com/2008/11/21/c-scripting-in-opensim/
[9:26]	Si Linden:	so one suggestion we have considered for modelling state is the State pattern: http://en.wikipedia.org/wiki/State_pattern
[9:26]	Cale Flanagan:	(go vote! :) )
[9:27]	Si Linden:	but as I say, it's still really up in the air, so nothing concrete yet!
[9:28]	Si Linden:	ok folks, I have to get back to interpreting my performance graphs
[9:29]	Si Linden:	I am going to ask babbage to look at the following for next week:
[9:29]	Si Linden:	Intergrid comms
[9:29]	Cale Flanagan:	nono, u justgot a worklist:)
[9:29]	Si Linden:	lol
[9:29]	Siann Beck:	lol
[9:29]	Si Linden:	Viewer bandwidth sharing
[9:29]	Nock Forager:	Thanks Si see you again.
[9:29]	Si Linden:	recompile in inventory
[9:29]	WolfPup Lowenhar:	tc
[9:29]	Si Linden:	angry kittens
[9:29]	Siann Beck:	heh!
[9:29]	Si Linden:	and open source plans for lsl -> mono
[9:30]	Si Linden:	:)
[9:30]	Siann Beck:	Cale wanted to know if you have a bear.
[9:30]	Cale Flanagan:	u firgit state loclas:)
[9:30]	Morgaine Dinova:	Hehe. Take care Si
[9:30]	Si Linden:	it's been great to chat to you all, see ya
[9:30]	Morgaine Dinova:	Si has a bear?
[9:30]	Si Linden:	state locals added to list
[9:30]	Darwin Recreant:	take care Si, ty
[9:30]	Cale Flanagan:	ah, he was nice, so he can leave without bear:) cu
[9:30]	Si Linden:	sorry no bear yet
[9:30]	Si Linden:	thanks Cale :)
[9:30]	Morgaine Dinova:	Hehe
[9:30]	Siann Beck:	:)
[9:30]	Si Linden:	byeeee!
[9:31]	Siann Beck:	Bye!
[9:31]	Morgaine Dinova:	Thanks for chat Si :-)
[9:31]	Fake Fitzgerald:	thanks and bye
[9:31]	Siann Beck:	Yes, thanks for coming!
[9:31]	Cale Flanagan:	now back to the 64k exploding storage lists..
[9:31]	Morgaine Dinova:	Was interesting.
[9:32]	Cale Flanagan:	bye all (and vote for state locels;) )
[9:32]	Morgaine Dinova:	I wonder how many scripting languages Opensim will have by next year. 20?
[9:32]	Imaze Rhiano:	42