User:Babbage Linden/Office Hours/2010 02 17

Transcript of Babbage Linden's office hours:

[3:04]	Sahkolihaa Contepomi:	Hey babbage.
[3:04]	Qie Niangao:	greetings, Babbage
[3:04]	Nock Forager:	Hello Babbage
[3:04]	ANSI Soderstrom:	lenovo business series have a trackPoint
[3:04]	Babbage Linden:	morning all
[3:04]	Aimee Linden:	but yeah, there's a discoverability issue with the mousewheel zoom too
[3:04]	ANSI Soderstrom:	the best mouse aalternative i ever seen
[3:04]	Aimee Linden:	hi Babbage :)
[3:04]	Ceawlin Steamweaver:	Mornin' Babbage
[3:04]	Imaze Rhiano:	good day
[3:04]	Babbage Linden:	how is everyone today?
[3:05]	Kaluura Boa:	Haaaa! AT last some good news... Let's hope
[3:05]	Ceawlin Steamweaver:	Wow, lots of people showed up while I was afk. O_O
[3:05]	Void Singer:	tired and cranky? =/
[3:05]	Sahkolihaa Contepomi:	Pretty good. New Snowglobe build I have is working beautifully. :)
[3:05]	Babbage Linden:	sorry to hear that void
[3:05]	Void Singer:	I'll get over it, but thanks
[3:05]	Babbage Linden:	so, let's get an agenda together, what would people like to talk about today?
[3:05]	Void Singer:	versioning
[3:05]	Ardy Lay:	Yes
[3:06]	Sahkolihaa Contepomi:	What broke the last sim update? :p
[3:06]	Kaluura Boa:	Yep
[3:06]	Ceawlin Steamweaver:	Efficient scripts on aditi, if they are not there yet. My usual spammery. XD
[3:06]	Imaze Rhiano:	versioning, how long you are going to support old versions
[3:07]	Babbage Linden:	so, last sim update, efficient scripts, versioning
[3:07]	Babbage Linden:	anything else?
[3:07]	Liisa Runo:	you need couple lights to your office, it is very dark
[3:07]	Babbage Linden:	i just turn the sun up liisa
[3:07]	Ceawlin Steamweaver:	I'm always curious about C# and Mono if there is time, but I am more interested in the other stuff at the moment. :P
[3:07]	Liisa Runo:	that is noob
[3:07]	Babbage Linden:	but, yes, good idea
[3:07]	Gunter Vandyke:	hi
[3:07]	Imaze Rhiano:	ya... and chair that is scripted such way that only you can sit on it :P
[3:07]	Simon Kline:	shift - ctrl - Y lisa :D
[3:07]	Questar Utu:	A throne of sorts?
[3:08]	Babbage Linden:	if anyone can find some lovely steampunk lights, i'll happily buy some and set them up here
[3:08]	Pete Olihenge:	More seats too..
[3:08]	Void Singer:	I'll bring ya some next time I come by....
[3:08]	Identity Euler:	just a sec I will find one
[3:08]	Qie Niangao:	(of course a scripter's meeting would have no facelights. :p )
[3:08]	Simon Kline:	we're going to need a grand stand at this rate hehe
[3:08]	Void Singer:	lol Qie
[3:08]	Questar Utu:	I could probably -make- some lights if I knew a particualr design to work around. <. <
[3:09]	Babbage Linden:	ok, that's 4 agenda items, let's start working through them and see where we get to
[3:09]	Void Singer:	gas lamp ok?
[3:10]	Babbage Linden:	so, the 2 issues that I'm aware of causing problems in the last release are
[3:10]	Babbage Linden:	https://jira.secondlife.com/browse/SVC-914
[3:10]	Ardy Lay:	https://blogs.secondlife.com/message/93928#93928
[3:11]	Babbage Linden:	and another that doesn't have an external JIRA
[3:11]	Ceawlin Steamweaver:	Ooooh souper sekret SEC jiras. O_O
[3:11]	Babbage Linden:	so, it was a well intentioned change that had some unintended consequences
[3:12]	Void Singer:	ya mean it exposed stupid logic bugs... =)
[3:12]	Babbage Linden:	since the incident we have set up an internal team to review all LSL changes
[3:12]	Babbage Linden:	void, no trolling please
[3:13]	Babbage Linden:	we got pretty good at understanding the consequences of LSL changes while we were working on mono
[3:13]	Void Singer:	sry, I only meant the referenced code...
[3:13]	Babbage Linden:	but we've hired a number of new developers since then
[3:13]	Babbage Linden:	and so need to spread the knowledge of how to work with LSL across the lab
[3:13]	Babbage Linden:	so, that's what we're doing
[3:13]	Babbage Linden:	the new team will review all changes
[3:13]	Qie Niangao:	how would a review detect this problem, though? could you really anticipate that there'd be so much code dependent on this bug?
[3:15]	Babbage Linden:	well, hopefully the team would have recognised this consequence
[3:15]	Babbage Linden:	when we were working on mono we tried hard not to change behaviour silently
[3:15]	Babbage Linden:	triggering a new event without changing the scirpt is new behaviour
[3:15]	Qie Niangao:	okay. well, to be honest, I like the "review" approach better than any alternative that comes to mind.
[3:16]	Babbage Linden:	this change wouldn't have met those criteria
[3:16]	Sasun Steinbeck:	Then again you could say the pilot rollout was successful in that it exposed a potential grid-wide problem, right? So that worked out well.
[3:16]	Babbage Linden:	as cg says on the blog we really need versioning here
[3:16]	Squirrel Wood:	checking for the appropriate event is good scripting practice. not checking = self owned
[3:16]	Babbage Linden:	so, changes that silently change behaviour would be made to a new version of the library
[3:17]	Babbage Linden:	new scripts could choose to use them
[3:17]	Babbage Linden:	old scripts would continue to use the old version of the library
[3:17]	Void Singer:	will we be on hold for LSL fixes pending versioning?
[3:17]	Babbage Linden:	void, not nescecarily
[3:17]	Ceawlin Steamweaver:	What happens when there are 5000 scripts in a sim all using 5000 different libraries though? That sounds like massive memory wastage to me. >_>
[3:17]	Babbage Linden:	silent, non-critical bug fixes will be on hold
[3:18]	Babbage Linden:	sometimes we have to make changes to close security issues
[3:18]	Babbage Linden:	it's a matter of understanding the consequences
[3:18]	Void Singer:	can you give an example of the type that would be on hold?
[3:18]	Nya Raymaker:	think it will more break down into "LSL before" and "LSL after".. hoping the new changes will be made consistent with each other
[3:18]	Babbage Linden:	now, versioned libraries are actually supported by Mono in the simulator right now
[3:19]	Babbage Linden:	the correct way to make this fix would have been to version the library and make all new scripts link against the new version
[3:19]	Babbage Linden:	but, that only works for Mono scripts
[3:19]	Babbage Linden:	which is still OK, we should have said, if you want this new, fixed behviour, recompile your scripts to this new version of the library
[3:20]	Babbage Linden:	or rather, recompile, which will now link your scripts to the new library
[3:20]	Babbage Linden:	then it is no longer a silent change
[3:20]	Babbage Linden:	you have to recompile and opt-in to the new behaviour
[3:20]	Ceawlin Steamweaver:	One thing that I have often wished for is a better way to be able to update scripts in-place on deployed objects, preferably without object owner interaction. As things currently stand, when one wants to update a product, the only options are to deploy an entirely new product, or have a +mod product and deliver an updater widget. I'd be 100% cool with behavior changes, if only there was an easy way to fix all the objects that are bouncing around there in the world after the fact.
[3:20]	Babbage Linden:	all other scripts work as they have been
[3:21]	Babbage Linden:	i will make sure this gets communicated internally
[3:21]	Void Singer:	that part sounds lovely... no more separate updates to script calls and legacy mess... how long are we talking about?
[3:21]	Babbage Linden:	we can do it now, i will send an email round
[3:21]	Babbage Linden:	we will want to batch fixes to avoid the 5000 libraries issue mentioned
[3:21]	Babbage Linden:	but, a new LSL library version every server release would be OK
[3:22]	Sasun Steinbeck:	do you understand the effects on sim performance to have, say 20-30 versions of the library in use on a sim?
[3:22]	Imaze Rhiano:	as Ceawling already said - versioning raises another problem - how long to support old versions? we can't support them forever, because then we would evenutually have hundreds different versions around.
[3:22]	Babbage Linden:	sasun, it would be the equivalent of having another 20-30 scripts running
[3:22]	Sasun Steinbeck:	I would love to know that scenario has been tested
[3:22]	Void Singer:	will there be ways to force legacy libraries? or detct their (lack of )usage for retirement
[3:23]	Babbage Linden:	we can scan assemblies for dependancies
[3:23]	Babbage Linden:	so we could fire up warnings saying that a script is using an old version of the library pending deprecation
[3:24]	Qie Niangao:	so... if some user has a mod object and recompiles an embedded script, they'll automatically get the newest library version? (or how will scripts specify which version they're to compile with?)
[3:24]	Ceawlin Steamweaver:	Yea, I personally don't think it will be feasable to support angient libraries into the far future. There's a point where some old content has to just be broken, or things will get all crufty and out of hand in the codebase, I'd think. >_>
[3:24]	Ceawlin Steamweaver:	*ancient
[3:24]	Babbage Linden:	qie, at the moment, without UI changed scripts would compile against the latest whenever recompiled
[3:25]	Babbage Linden:	which is probably the easiest policy
[3:25]	Ceawlin Steamweaver:	I am cool with that and all, don't get me wrong. I think it's gotta be done eventually. I just wish there was a better way to deal with that kinda stuff on our side. :P
[3:25]	Babbage Linden:	i agree with the discussion of auto updating
[3:25]	Babbage Linden:	it's very hard to update now
[3:25]	Babbage Linden:	but a much bigger project
[3:26]	Sasun Steinbeck:	yes, amen to auto updating!
[3:26]	Babbage Linden:	and also agree that we'd need to have some policy about retiring old versions of the library
[3:26]	Babbage Linden:	these issues will come up in the C# development too
[3:26]	Squirrel Wood:	force-recompile on rez when you are n versions behind ?
[3:27]	Pete Olihenge:	How will this affect the timing of other expected critical fixes?
[3:27]	Babbage Linden:	it would be good to be able to specify library versions there
[3:27]	Gunter Vandyke:	what about a llRecompile() function
[3:27]	Babbage Linden:	pete, normally if something is a critical fix, we just make the change
[3:27]	Ceawlin Steamweaver:	I'd like to have a script able to create and compile other scripts after fetching some source code (or even an assembly) from some place or other. HTTP perhaps. ;P
[3:27]	Babbage Linden:	if something is crashing the sim, it's OK to fix it
[3:27]	Squirrel Wood:	I hope C# will allow me to finally use a proper PRNG
[3:27]	Babbage Linden:	as the sim would crash anyway
[3:27]	Void Singer:	last biggie question for me... what types of fixes (I already assume lso ones) will likely be on hold? I can assume not SEC or similar
[3:28]	Babbage Linden:	it's more difficult to fix something that is broken but relied on
[3:28]	Questar Utu:	.. llRecompile()? Talk about heavy load on the sim if used wrong..
[3:28]	Babbage Linden:	a bug that has become a feature
[3:28]	Babbage Linden:	I think manual recompilation is better
[3:29]	Squirrel Wood:	llRequestCompile?
[3:29]	Babbage Linden:	having a scripting actually able to see errors and test output is a good thing ;-)
[3:29]	Void Singer:	gotcha, so changes to things like posJump are off the list for now...
[3:29]	Squirrel Wood:	with the user having to OK it ?
[3:29]	Void Singer:	that's actually happy news
[3:29]	Babbage Linden:	I wouldn't want to automatically recompile my scripts live and just hope they worked ;-)
[3:29]	Babbage Linden:	but I would like to be able to recompile a script, test it and then say, this is a new version of script foo, distribute it
[3:30]	Babbage Linden:	or have an llCheckForUpdate("MyScript");
[3:30]	CPU Core:	I dont see no need for recomipling each script.. they just need a better way to replace compiled scripts
[3:30]	Babbage Linden:	or something
[3:30]	Babbage Linden:	right
[3:30]	Squirrel Wood:	these days recompiling creates a new asset anyway..
[3:30]	Babbage Linden:	ok, we're half way through the hour and have done 1 topic
[3:30]	Squirrel Wood:	like editing a notecard does..
[3:30]	Babbage Linden:	let's move on
[3:30]	Ceawlin Steamweaver:	XD
[3:31]	Babbage Linden:	good discussion though
[3:31]	Babbage Linden:	thanks
[3:31]	Babbage Linden:	so, next up we had efficient scripts status
[3:31]	Babbage Linden:	it's still on course for 1.38
[3:31]	Babbage Linden:	and kelly has been pulling in some of the nice to have features we had on the backlog
[3:32]	Ceawlin Steamweaver:	Sorry but, I have to ask..... What's the latest guesstimate of the date it'll be on ye ole beta grid, do you know? :P
[3:32]	Questar Utu thinks it would be nice to see script time even as a non-sim owner/EM, to test script efficiency.
[3:32]	Squirrel Wood:	make sure you add sonething in so everyone can check their *own* script run times and such
[3:32]	ANSI Soderstrom:	what about version and compilling controlling with the already well known method like w3c (<!DOCTYPE lsl PUBLIC "-//LL//DTD XLSL 1.0 Strict//EN" "http://Sl-Page/-strict.dtd">)
[3:32]	Babbage Linden:	llLinkParticlesystem is now done
[3:32]	Babbage Linden:	and llSetLinkTextureAnim
[3:32]	Ceawlin Steamweaver:	Ooo cool.
[3:32]	Nya Raymaker:	oh cool
[3:32]	Qie Niangao:	oh, superb!
[3:33]	Kaluura Boa:	llSetLinkText()?
[3:33]	Babbage Linden:	and LRIM_TEXT
[3:33]	Kaluura Boa:	Goood!
[3:33]	Qie Niangao:	*wow*
[3:33]	Questar Utu:	.. Link particles.. Hooh boy that's not going to turn out well..
[3:33]	Babbage Linden:	PRIM_TEXT, even
[3:33]	Sasun Steinbeck:	\o/
[3:33]	Nya Raymaker grins happily
[3:33]	Becky Pippen:	wow this is like christmas
[3:33]	Babbage Linden:	all of these things are doable now, but require lots of scripts
[3:33]	Squirrel Wood:	llSetLinkPrimitiveParams([PRIM_TEXT,"",<0,0,0>,1]); ?
[3:33]	Squirrel Wood:	or something like that?
[3:34]	Babbage Linden:	llSetLinkPrimitiveParamsFast([PRIM_TEXT,"",<0,0,0>,1]); yes ;-)
[3:34]	ANSI Soderstrom:	llLinkParticlesystem has already some erros... isnt done
[3:34]	Ceawlin Steamweaver:	Still aiming for sometime this for testification of that? Or is it already on aditi and I missed it? >_>
[3:34]	Ceawlin Steamweaver:	*this month
[3:34]	Void Singer:	net get's for all the exclusive sets, and set's for all thexclusive gets.... but I'm sure you've already heard that complaint ;) (exception for script pin)
[3:34]	Object: Hello, Avatar!
[3:34]	Babbage Linden:	i will find out what the beta plan date is
[3:35]	Ceawlin Steamweaver:	Thankies. ^_^
[3:35]	Kaluura Boa:	Seriously? llSetLinkPrimitiveParamsFast()? Couldn't you find a longer name?
[3:35]	Babbage Linden:	once it's through QAR it will go on to the beat grid
[3:35]	Squirrel Wood:	any way to set text on a prim in a linkset where the prim has a distinct name?
[3:35]	Babbage Linden:	squirrel, no it's all by link number
[3:35]	Nya Raymaker:	how about the equivalents to actually get these link prim parameters?
[3:35]	Squirrel Wood:	aww.
[3:35]	Babbage Linden:	you could set up a mapping though
[3:35]	Questar Utu:	Change it to an acronym! :D llSLPPF()
[3:35]	Aimee Linden:	yes, llSetLinkPrimitiveParamsFasterThanTheOldOne
[3:35]	Void Singer:	eh, I can live with that
[3:35]	Squirrel Wood:	link numbers tend to get jumbled up when unlinking and relinking
[3:35]	Imaze Rhiano:	lol, Aimee :P
[3:35]	Ceawlin Steamweaver:	Squirrel: I usually scan the link set and generate a hash map for that at init time.
[3:36]	Sahkolihaa Contepomi:	Aimee, don't give ideas. :p
[3:36]	Babbage Linden:	suc(name) -> return name + "MoreOssm!!!i!!"
[3:36]	Sasun Steinbeck:	LOL
[3:36]	Sahkolihaa Contepomi:	o_O
[3:36]	Nya Raymaker:	that is why gettign the link params is something I'd like
[3:36]	Squirrel Wood:	ya. I generate a map too but would like to see that memory waster gone :p
[3:36]	Babbage Linden:	nya, yes there is a GetLinkPrimitiveParams too
[3:36]	CPU Core:	why dont they just use the old name.. its not like it would break anything and in most cases it would make peoples creations work better without them having to do anything
[3:36]	Ceawlin Steamweaver:	Make a hash map instead of a string map. Fast searches, little memory use. ;P
[3:36]	Void Singer:	out of curiosity, will these eventually be tied into versioning
[3:37]	Squirrel Wood:	llSetLinkPrimitiveParamsFasterThanLight :p
[3:37]	Nya Raymaker:	so you still have somethign to check back on.. ah okay ;)
[3:37]	Babbage Linden:	ceawlin, use a System.Collections.Hashmap...
[3:37]	Babbage Linden:	ok, so that's where efficient scripts is, I will check the beta date
[3:37]	Ceawlin Steamweaver:	I am waiting hungrily for the day, Babbage. XD
[3:37]	Aimee Linden:	llSetLinkPrimitiveParamsFasterThanYouCanTypeIt, I think was my favourite
[3:37]	Squirrel Wood:	llFindLinkPrimNamed("prim_name") whch returns the link number?
[3:37]	Babbage Linden:	next, is versioning, which we covered a lot in the sim update discussion I think
[3:38]	Babbage Linden:	anything more we should discuss there?
[3:38]	Imaze Rhiano:	.. and memory limit UIs?
[3:38]	Nya Raymaker:	sort of....
[3:38]	Jonathan Yap:	I think we now need a way to build gestures that will go into the edit window, so you can program keys to spit out these long names
[3:38]	ANSI Soderstrom:	llGiveback();
[3:38]	Nya Raymaker:	the versioning wil not only be an issue for people who actually can write a script
[3:38]	Squirrel Wood:	I want a commemorative rolling-restart bear!
[3:38]	Nya Raymaker:	but if a version is retired that is being used in stuff out there... then fashion addict x suddenly has stuff that does not work...
[3:38]	Babbage Linden:	by versioning were we talking about script versioning? or auto-updating?
[3:39]	CPU Core:	those who buy full mod scripts will have problems :)
[3:39]	Babbage Linden:	so, any auto-updater needs to be optional
[3:39]	Nya Raymaker:	and sending them a "your script xyz has been retired" message will.. cause wide spread confusion
[3:39]	Babbage Linden:	"Your thing will stop working when LL deprecates script library 2.5 would you like to update..."
[3:39]	Squirrel Wood:	your script xyz is too old. please obtain a more recent version
[3:40]	Ceawlin Steamweaver:	I'd prefer to just see some stuff in the API that we can use to write our own updaters appropriate to the situation, rather than being locked into some particular LL provided system, no offense. ;P
[3:40]	Void Singer:	'is broken, buy new one"
[3:40]	ANSI Soderstrom:	"install opensim to see the script working"
[3:40]	Squirrel Wood:	your animation override script suffers from the same error most other ao scripts suffer.. missing animations cause script error spam :p
[3:40]	Imaze Rhiano:	How are memory limits coming together?
[3:40]	Babbage Linden:	right, well you can write scripts that phone home, find out about updates, request them then delete themselves now, it's just painful
[3:41]	Ceawlin Steamweaver:	Yea, but they all require owner interaction to rez a delivered updater, and for the object being updated to be +mod.
[3:41]	Nya Raymaker:	yup
[3:42]	Babbage Linden:	right, so that's what we're talking about fixing here
[3:42]	Cerdita Piek:	May I introduce a new topic? Any news about SVC-3895? how is it progressing?
[3:42]	Ceawlin Steamweaver:	I've tried to convince various clients that making their objects -mod doesn't do anything to stop piracy, but a lot of them still like to -mod their stuff, and then it is hell time when I have to update stuff. X_x
[3:42]	Void Singer:	will these pretty new fast functions with extremely long names be replacing the old built in delayed ones under the versioning sytem? it'd certainly ease teaching overhead
[3:42]	Nya Raymaker:	basically.. this will run down to having even the smallest item delivered with a proof of purchase so people can get updates from a vendor server if soemthign gets abandoned
[3:42]	Babbage Linden:	void, it's certainly an option...
[3:43]	Babbage Linden:	discussion to continue
[3:43]	Babbage Linden:	right let's quickly finish the agenda, then we can add new stuff
[3:43]	Babbage Linden:	C# and mono
[3:43]	Babbage Linden:	we're just starting a new sprint now to add the moonlight coreclr security features to the simulator
[3:43]	Jonathan Yap:	Cerdita, look at 3:34 here http://wiki.secondlife.com/wiki/User:Babbage_Linden/Office_Hours/2010_02_10
[3:44]	Ceawlin Steamweaver:	Ooooo cool
[3:44]	Babbage Linden:	that's the big thing that's stopping us have a resident beta
[3:44]	Cerdita Piek:	thanks, jonathan
[3:44]	Babbage Linden:	so we want to get that done soon so we can get feedback as soon as possible
[3:44]	Babbage Linden:	we're going to be updating to Mono 2.6 as part of this too
[3:44]	Babbage Linden:	as that is going to have long term support
[3:45]	Ceawlin Steamweaver:	What's your guesstimate on the amount of time from beta deployment to main grid deployment once that is done?
[3:45]	Babbage Linden:	ceawlin, I expect a fairly long beta period, both to fix bugs and iterate on the API
[3:45]	Babbage Linden:	when we developed Mono we had a 6 month beta
[3:45]	Ceawlin Steamweaver:	I ask cos I will be tasked with updating a bunch of stuff to C# when it goes on Aditi, and they will want to know when it will be usable on maingrid, lol.
[3:45]	Babbage Linden:	and I would expect that to be the case here
[3:46]	Ceawlin Steamweaver:	Okies.
[3:46]	Babbage Linden:	we want to get the api right this time ;-0
[3:46]	Ceawlin Steamweaver:	Hehehe
[3:46]	Babbage Linden:	ok, so let me find the other agenda items.
[3:46]	Imaze Rhiano:	Memory limitis and mono rezzing problem?
[3:46]	Babbage Linden:	SVC-3895
[3:47]	Babbage Linden:	right
[3:47]	Babbage Linden:	memory limits is very nearly done
[3:47]	Babbage Linden:	we're fixing one last server issue
[3:47]	Void Singer:	and sim crossings for the same if it's related
[3:47]	Babbage Linden:	and the server side should be in 1.38#
[3:47]	Squirrel Wood:	mono DE-rezzing problem ? ^^
[3:47]	Babbage Linden:	once the server side is released it will grant caps to the new viewer
[3:47]	Babbage Linden:	which will show the UI
[3:47]	Squirrel Wood:	I can still make sims cry out in pain by simply derezzing my terraformer :p
[3:47]	Ceawlin Steamweaver:	Ooooo, will that include memory > 64k at the same time? O_O
[3:48]	Babbage Linden:	nothing will be enforced, but it will show usage
[3:48]	Babbage Linden:	LSL scripts will be 16K, Mono scripts will be 64K
[3:48]	Babbage Linden:	we're working on the dependencies for the next step
[3:48]	Sahkolihaa Contepomi is glad he'll no longer have to keep looking at the top scripts window in the EM panel for his own script time.
[3:48]	Babbage Linden:	I think we need to add support for small scripts
[3:48]	Ceawlin Steamweaver:	Ahh ok. :P The only thing I want more than efficient scripts right now is to be able to request heap sizes > 64k, lol.
[3:48]	Babbage Linden:	Mono scripts asking for <64K
[3:48]	Babbage Linden:	and bytecdoe sharing
[3:49]	Babbage Linden:	before we enforce script limits
[3:49]	Void Singer:	good to hear
[3:49]	Babbage Linden:	and then allow Mono scripts to ask for > 64K
[3:49]	Babbage Linden:	it's going to be a long process
[3:49]	Babbage Linden:	as we also want to allow feedback on the proposed limits
[3:49]	Squirrel Wood:	conditional compile options in mono scripts ?
[3:49]	Nya Raymaker:	but yes.. that allows slave scripts to run low level.
[3:50]	Qie Niangao:	oh: crediting for shared bytecode... that's nice.
[3:50]	Babbage Linden:	so, last topic SVC-3895
[3:50]	Babbage Linden:	I've made 2 further improvements
[3:50]	Babbage Linden:	I now look for assemblies in all Mono Application Domains before loading a script
[3:50]	Babbage Linden:	which should completely avoid the loading scripts twice delays we were seeing
[3:51]	Babbage Linden:	I've also made some changes that make scripts lazily create themselves
[3:51]	Squirrel Wood:	Babbage, derezzing scripted objects causes even worse sim performance drop ^^ and I can prove it.
[3:51]	Babbage Linden:	for example, the creation process for a script created a script object
[3:51]	Babbage Linden:	and then deserialized it
[3:51]	Babbage Linden:	a mono object was created for the script
[3:52]	Babbage Linden:	and then immediately discarded when the script was deserialized
[3:52]	Babbage Linden:	now the creation call doesn't create a managed object
[3:52]	Babbage Linden:	so it's not wasted
[3:52]	Babbage Linden:	we wait until we deserialize or run the script to create the mono object
[3:52]	Babbage Linden:	or otherwise call in to managed code
[3:53]	Babbage Linden:	there is still some work to be done on that, but it's nearly there
[3:53]	Babbage Linden:	and should be in 1.38
[3:53]	Babbage Linden:	also, Simon Linden has been working on multithreading rez and derez
[3:53]	Void Singer:	will that cause any script pauses for sim crossings?
[3:53]	Babbage Linden:	which would fix all of these problems
[3:54]	Babbage Linden:	but he has quite a lot of work to do there
[3:54]	Babbage Linden:	my changes should make rezzing Mono scripts much faster
[3:54]	Babbage Linden:	I would like to no about derezzing problems
[3:54]	Babbage Linden:	know, even
[3:54]	Babbage Linden:	is there a JIRA?
[3:54]	Squirrel Wood:	I can show you after the OH if you want
[3:55]	Babbage Linden:	I'd like a public JIRA with a minimal repro please :-D
[3:55]	Squirrel Wood:	no jira for it yet I think
[3:55]	Squirrel Wood:	I'll see what I can do
[3:55]	Babbage Linden:	thanks
[3:56]	Squirrel Wood:	though you may just get a video repro :p
[3:56]	Babbage Linden:	ok, we have 4 minutes left and have got through everything
[3:56]	Ceawlin Steamweaver:	\o/
[3:56]	Babbage Linden:	anything else?
[3:56]	Imaze Rhiano:	maybe slow derezzing is causing application domain switch
[3:56]	Jonathan Yap:	Are you still on a 6 week release cycle?
[3:56]	Ceawlin Steamweaver:	I wanna know then llDoWhatIAmThinking() will be done. O_O
[3:56]	Gunter Vandyke missed the date we get the new functions ㋡
[3:56]	Squirrel Wood:	llFixBrokenScript();
[3:56]	Babbage Linden:	Gunter, I'm going to find those dates for you
[3:57]	Sahkolihaa Contepomi:	llFixBrokenAnimationOverrider();
[3:57]	Nya Raymaker:	llRefillCoffeeCup is on my priority list ;)
[3:57]	Questar Utu:	Heh.
[3:57]	Gunter Vandyke smiles
[3:57]	Sahkolihaa Contepomi:	LOL.
[3:57]	Void Singer:	Ceawlin... isn't that just llReadScriptersMind?
[3:57]	Babbage Linden:	speaking of refilling coffee cups...
[3:57]	Ceawlin Steamweaver:	My clients want me to implement a library for llReadClientsMind(), but I can't get it to work. X_x
[3:58]	Squirrel Wood:	llStartParty();
[3:58]	Babbage Linden:	I'm going to be working on my SL Pro talk this week
[3:58]	ANSI Soderstrom:	first u must collect their karma
[3:58]	Void Singer:	Ceawlin... there may be a missing component in some case =X
[3:58]	Squirrel Wood:	llFindAndEnslaveDJ();
[3:58]	CPU Core:	what is SL Pro
[3:58]	Nya Raymaker:	that's cause you don't need a library there... but you need a giant black box to tap into
[3:58]	Babbage Linden:	and hopefully also getting some instructions done for the LSL OAuth library released
[3:58]	Ceawlin Steamweaver:	Yea, I think most of the problem is between my chair and my keyboard. ;<
[3:58]	Babbage Linden:	SL Pro is an in world conference for SL content creators
[3:59]	CPU Core:	cool
[3:59]	Jonathan Yap:	SL Pro info https://blogs.secondlife.com/community/commerce/blog/2009/12/17/sl-pro
[3:59]	CPU Core:	ty
[3:59]	Void Singer:	I was thinking the end users chair...
[3:59]	Babbage Linden:	Also, the video of my FOSDEM talk should be online soon
[3:59]	Babbage Linden:	So keep an eye out for that
[3:59]	Nya Raymaker:	:) well one advantage of versioning?
[3:59]	Babbage Linden:	it includes a couple of glimpses of the current C# API...
[3:59]	Questar Utu:	Hm.. An actual script function I just thought of that would be useful for some things. llInterpolateColors(vec1,vec2,stepcount,step);
[3:59]	Nya Raymaker:	whenever we thoguth we made soemthign fool proof we can rewrite it to the new version cause we met a bigger fool ;)
[4:00]	Babbage Linden:	ok, time's up!
[4:00]	Sasun Steinbeck:	when will 1.36 be pilot deployed again, and will it be deployed on the same pilot servers as last time?
[4:00]	Ceawlin Steamweaver:	Thanks for your time, Babbage! ^_^
[4:00]	Babbage Linden:	thanks for coming everyone
[4:00]	Ardy Lay:	Babbage, I mentioned derez and gave you objects in October SVC-3895
[4:00]	Cerdita Piek:	Thank you, Babbage. Take care :)
[4:00]	Becky Pippen:	thanks Babbage
[4:00]	Sahkolihaa Contepomi:	Thanks for the talk, Babbage.
[4:00]	CPU Core:	bye
[4:00]	Qie Niangao:	Thanks, Babbage.
[4:01]	Void Singer:	Nya after breaking a sim in two line of code on accident... I could be that bigger fool
[4:01]	Imaze Rhiano:	thanks Babbie
[4:01]	Babbage Linden:	good discussion today
[4:01]	Gunter Vandyke:	thanks babbage
[4:01]	Sebastean Steamweaver:	ARg!
[4:01]	Void Singer:	thanks babbage