Viewer Evolution User Group/Archive/2011-04-15

08:02	draconis.neurocam	hello oz
08:03	Oz Linden	Hi all
		Agenda is here: https://wiki.secondlife.com/wiki/Viewer_Evolution_User_Group
	Haravikk	So open development now equates to closing issues that shouldn't be?
	Oz Linden	Open development does not mean that Linden Lab does not get the final call
08:04	Haravikk	You haven't given a single reason why VWR-7830 should legitimately be closed other than you say so
	Oz Linden	it also means that if you want it done another way, you can build a TPV that does
		That issue isn't on the agenda....
		at least not at the start of it
08:05	Haravikk	Just because you're opting for a minor, only barely related "solution" in the mean time doesn't mean there is no future viability to VWR-7830
	Oz Linden	VWR-24731
		hmm... helper busted?
	Haravikk	I dunno, maybe I'm biased towards discussing it now because if it doesn't fit the meeting's time limit then I have no JIRA priviledges for a week because you'd rather resort to bully tactics than give clear reasoning
08:06	Oz Linden	well, wait
		GPU table updates
		I'm in the process of doing this now - I've merged the updates that the Phoenix project added, and have a couple of assorted others to plug in
08:07		There were lots of errors (ordering problems) in the table from Phoenix, and some in our copy
	Tonya Souther	What kind of ordering problems?
08:08	Oz Linden	I'd love to have someone develop a unit test for that recognition.... if someone is looking for a project that would be good for everyones graphics, this is a good one that doesn't require anything more than basic c++ skills
		The table is searched linearly and first match stops the search
		there were lines that could never be matched because an earlier expression would find them first
08:09		I don't think any actually had the wrong class as a result, but they would have gotten the wrong name
		things like matching "23..*" first, followed by "23M.*"
	draconis.neurocam	which would lead to improper statistical data?
08:10	Oz Linden	I'm not sure... possibly.... I have not figured out yet whether the stats are based on the real name or on the recognized name
		I'm looking into that, and trying to find a utility that someone allegedly once built that would run the stats data against the viewer table to figure out what is not recognized
08:11		if I can't find the old utility, it won't be hard to recreate
08:12		So ... if you know of any GPU recognition issues that are not currently linked to STORM-1100, now is a good time to link them
	JIRA-helper	http://jira.secondlife.com/browse/STORM-1100 STORM-1100 Add new and changes to GPU_Table.txt
	Oz Linden	ah.... the helper was fooled by the move... have to look into that....
		next update ... Windlight Estate....
08:14		we're still working on this ... been doing lots of QA, and have a UI review tomorrow. We're going to do some rework on the UI for this, I think, and the project will most likely extend into the next Snowstorm sprint.
		but we're making real progress
	draconis.neurocam	good to know
	Oz Linden	will get a project viewer out sometime in the next week or two (and then you'd see how gloomy it is here right now :-) )
08:15	Tomkin Euler	? but...
		:)
	Oz Linden	VWR-24764
	JIRA-helper	http://jira.secondlife.com/browse/VWR-24764 VWR-24764 Make parcel_media_agent_command and similar request user permission
08:16	Oz Linden	not much news on that one...
		Tonya - it's failing a unit test, so we need to figure that out
		will get you details after we're done here
	Tonya Souther	is the media filter in there?
08:17	Oz Linden	I have it in a test repo, but it doesn't build cleanly
	Tonya Souther	I'll get wiht you about that, then.
08:19	Oz Linden	So... the parcel permissions thing.... I don't want to have a debate about process here & now (will be happy to have it another time). We concluded that modifying the active group has too many different unrelated effects to do it automatically. Arguably, that is a flaw in the 'active group' mechanism, but if so that's a much larger issue with lots of backward-compatibility ramifications.
08:20	Haravikk	That's still not a reason to close VWR-7830, just because some other solution is being considered in the short term has no effect on the future viability of VWR-7830 as it still ultimately the ideal solution since it actually fixes the problem
	Oz Linden	So we're going to come at the parcel permissions issues in other ways (possibly including some other than STORM-313)
	JIRA-helper	http://jira.secondlife.com/browse/STORM-313 STORM-313 Build parcel property is displayed incorrectly
08:21	Haravikk	Closing an issue implies that it has no possibility of ever being considered, which is simply not true, however complex it may or may not be in practise
08:22	Oz Linden	Like I said, I don't want to have a process discussion, but I'll have just this much of it: the Jira is not a debating forum, it's a work tracking system. Once we've decided something is not a good description of work we would like to be done, we'll close the issue. VWR is a horrible almost unusable mess, and it's going to get cleaned up. Thats going to mean closing a _lot_ of issues.
	Haravikk	It should only be closed ONCE the problem is solved in some other way, or if the problem is reasonably deemed to be unsolvable
08:23	Oz Linden	No, it's closed when we decide the issue is no longer useful to us.
	Tonya Souther	Oz, is there another way Haravikk can keep the issue on LL's radar?
	Oz Linden	I am not asserting that there is not a problem
		Quite the opposite- there are a number of problems, but that issue is not the solution and is no longer useful
08:24	Haravikk	Until that problem is solved VWR-7830 is a viable solution that users should be able to vote on, comment about and show support for
	Oz Linden	We disagree about how Jira should be used, Haravikk - but it's our jira, so ....
08:25	Haravikk	Where is this "our"? The whole point of the public JIRA is for people to vote on issues to show support for them and hopefully get them prioritised, if it has public access then it's not just your JIRA
	Tomkin Euler	thre is not an uber issue it could fit in? yet?
08:26	Oz Linden	There are an almost infinite number of ways we could possibly address any given problem... if we allowed all related issues to remain open for every unsolved problem, we would have far too many to be able to find anythign
	Tomkin Euler	:0)
	draconis.neurocam	which is how it has been for a long time, and its good it is being cleaned >.>
	Oz Linden	I think I've made my position both on the issue and on Jira usage clear.... let's move on.
08:27		VWR-25506
	JIRA-helper	http://jira.secondlife.com/browse/VWR-25506 VWR-25506 Investigate restrictions on storing and exchanging inventory items within prims and notecards
	draconis.neurocam	yes i know i offered help oz, but i cant find the wiki page mentioned in the article to test it
	Haravikk	This is bollocks, I've been a resident for over 5 years, you've been a Linden for what, nearly one? The reisdents are supposed to have a say in how SL is improved and you seem content to spit on that and call it open development?
08:28	Oz Linden	Haravikk... if you'd like to discuss that with me further, I'll be glad to set up another time to do that
	Haravikk	improved parcel permissions display isn't a fix at all, and yet you want to close the only other altenrative currently available
08:29		What, so you can do it away from other people? You've already revoked my bloody ability to post on the JIRA, I have no respect for you in the slightest anymore
		I'll be taking this up with other Lindens instead if you're unwilling to discuss this in public
08:30	Oz Linden	I'll see if I can find the KB page that describes those asset moving restrictions and send them to you, draconis
	draconis.neurocam	alright,
	Tomkin Euler	no viewer specified.. any will be ok to use?
		development?
08:31	Oz Linden	I am pretty sure that at least most of the restrictions can be either eliminated or at least made unreasonably large (an order-of-magnitude or two larger)
	draconis.neurocam	i would agree
	Haravikk	To hell with it, more and more of these sessions are becoming breeding grounds for dismal of valid issues rather than places for solutions, you're making a mockery of what Esbee was doing, it's obvious there is no place for people who care about SL it seems, of which there are only going to be less and less as you continue to alienate everyone
	Oz Linden	I don't think those limits have been changed in years, so any reasonably recent version is relevant
08:32	oobscure	I plan to do testing on this issue
		during the weekend
08:34	Oz Linden	It's possible that we'll end up with one set of limits if the viewer is using HTTP and a different one if UDP, but in the long run we'll only use HTTP and the larger limits would always apply
	Tonya Souther	Not all HTTP limits are larger.
	Tomkin Euler	Haavikk cares about his issue.. :/
	Oz Linden	There's some additional instrumentation on that in the pipeline that should be integrated into viewer-development soon....
08:35		will reach out to Haravikk privately ... I care about the issue too
	Tomkin Euler	the jira does not refer to code sections?
	Oz Linden	code sections?
08:36	Tomkin Euler	where the operations are done, what the limits are..
		files in the tree..
	Oz Linden	oh... no, not yet
		I have not looked that far yet.
08:37		I want to get a handle first on 1) what the current and legacy viewers allow, and then 2) what should be allowed
		then we'll get to actually figuring out where the changes need to be made
	Tomkin Euler	k
08:38	Oz Linden	it's not unlikely that there are asset system implications
	draconis.neurocam	there should be some wiggle room despite that, i cant imagine the old limits are still valid, whatever they may be in the situations concerned
08:39	Oz Linden	once we've got the current limits properly characterized and tested, I'll get with the people working on the asset system (there's a lot going on there now) and figure out what if any changes are needed outside the viewer
	Techwolf Lupindo	I've been here 15 minutes and still havn't figure out what limit are being discussed now.
	draconis.neurocam	restrictions on storing and exchanging inventory items within prims and notecards
08:40		is the title of the jira
		anyway
	Oz Linden	How many objects can you move from your inventory at one time into something else (a prim, a notecard, .... )
	draconis.neurocam	i would imagine avatar to avatar should also be tested
	Oz Linden	yes, avatar to avatar should be looked at too, but...
		that has other implications
08:41		if you give something to another avatar, there are notices generated
	draconis.neurocam	right
	Oz Linden	I suspect that the real limit right now on that path is in the notice system
	oobscure	Techwolf: VWR-25506
	Oz Linden	which is another thing that the backend folks are spending some time on these days
	Techwolf Lupindo	Would love to dump everything I got years ago into a notecard and have an inventory that loads in seconds instead of hours.
08:42	Oz Linden	I got an interesting suggestion last night, actually, that I'd like your reaction to....
08:43	Tomkin Euler	calling cards will be restricted, no traansfer objects should be tested. how the container's permissions are changed slhold be tested..
	Oz Linden	The suggestion was that there be a web mechanism for accessing your full inventory, and that there be a distinction between your full inventory and what your avatar was 'carrying' at any one time.
08:44		The viewer inventory would show only what you were 'carrying'
	Dzonatas Sol	if by web we mean restful... =)
	Techwolf Lupindo	first level of folders or two would load in seconds.
	draconis.neurocam	would be interesting, but i dont even want to think about the effort needed to do, and the problems that would arise because of
	Oz Linden	The (presumably richer and more flexible) web UI to your full inventory would allow you to mark anything as either carried or not
08:45	draconis.neurocam	not to mention possible insecurities, if it allowed for better searching , by creator, description, etc im sure people would love it
08:46	Oz Linden	Re: RESTful, it's my personal positon that all services should have apis (and RESTful is a good api paradigm) and rich HTML versions.
	Techwolf Lupindo	If the web UI is done right, the number of support tickets of "missing inventory" should be less as one can verfiy that indeed the item is in there inventory.
	Tomkin Euler	I think the inworld inventory handles 9000 things very well.. and the web might not, realistically..
	Tonya Souther	If things not selected to be "carried" were simply not presented to your viewer aat login, it might not need any viewer changes at all.
	Dzonatas Sol	Thanks, Oz =)
08:47	Tomkin Euler	a carry-on bag for the basic viewer might be cool, tho..
	Techwolf Lupindo	The web restfull api should be configure to send only what the client needs, ie: don't send full details of 100K inventory items in one huge responce.
	Tomkin Euler	cloths, notes, landmarks.
08:48	Oz Linden	With respect to the effort... one of the motivators for moving some things to web interfaces is that it's _much_ easier to develop web UIs than to develop similarly powerful UIs within the viewer
		and it's far far easier to find developers with the requisite skills
	Techwolf Lupindo	Easer to hire cheap web devolopers then good c++ coders.
	Tomkin Euler	what performance difference do you sense, tho, Oz?
	Tonya Souther	Easier to *find* cheap web developers than even mediocre C++ coders.
08:49	Oz Linden	I make it a policy not to guess at performance, but if the viewer simply was never told about things that were not 'carried', then there is the potential (if the user were clever) for there to be far less data - it's unlikely that would be bad
08:50	draconis.neurocam	you have to assume the worst of people if you want to guess at performance, but yes, entirely possible
	Techwolf Lupindo	Might cause some annoyiouns. "I have to go to web page to carriy the outfit, then go back to the lcient to wear it"
08:51	Oz Linden	Well, even if the user just made everything 'carried', that just degenerates to exactly what we have now, right?
	draconis.neurocam	true
		so the benefits outweigh
		the real win is if it would offer better search capabilities
	Tonya Souther	Right. So make the default to carry everything...which suggests the metaphor would be more like closeets you put things in you don't need.
08:52	draconis.neurocam	that would make people who look for those search options in tpvs more likely to use the main client
08:53	Techwolf Lupindo	Thats hersity Draconis.
		:-)
	Tomkin Euler	these objects reside in the servers in the same way, tho?
	Oz Linden	Yes, that's essentially it... the person who sent it (Madeline BloodDoll - lovely name :-\ ) pointed out that it would eliminate the need to stash things in prims.
	draconis.neurocam	it would save tpv developers work to maintain there code relating to it as well tech
		their*
08:54	Tonya Souther	In fairness, I doubt that those who designed the system originally had any idea people would have inventories with 100K items and 50K in a single folder.
	Oz Linden	I wonder...
	draconis.neurocam	i have heard a linden or two say the most regretable thing about secondlife is not setting inventory limits
	Oz Linden	Not many have that much in single folders... we looked
	Techwolf Lupindo	Gridhunters have high inventory counts.
08:55	Tonya Souther	Unfortunately, two mf my RL friends do. :3
	Oz Linden	yes, well, there's always someone out on the long tails of the curve
	Tonya Souther	nods.
08:56	draconis.neurocam	imagines regexing every parameter of his inventory available
08:57	Techwolf Lupindo	Expecially finding that one megaprim size Drac
	draconis.neurocam	i use salt for that
	Oz Linden	the hardest part of that would probably be introducing the extra 'carried' meta-data :-(
	draconis.neurocam	hehe
		theres not already a good deal of that oz?
	Oz Linden	I don't know, really
08:58	Techwolf Lupindo	Oz, that can be rolled into the object metadata project. LL has said they wanted to add meta deata to objects to allow liciencesing.
	draconis.neurocam	ooh, better ip handling
		i bet that sounds good to a lot of lindens
08:59	Oz Linden	Well, I don't have any comment on that, except to say that the simplest way to kill any engineering effort is to keep adding requirements to it
	draconis.neurocam	we are just mentally expanding on the possibilities is all oz
09:00	Techwolf Lupindo	Not requirements. Sujest it to the team.
	Oz Linden	I read an article once about the development of the A-10 (a tank-killing airplane), in which the guy who managed it for the Air Force said that his main contribution was to file most of the suggestions that came from the various reviews and never tell the development team about them. :-)
09:01	Tomkin Euler	so, a first iteration of this idea is a prototype inventory interface from servers to the web.. through secondlife.com?
	Techwolf Lupindo	Just sujest to the team to make sure that its easer to add flags later down the road then to have to create a whole new project just to add that flag.
	Oz Linden	I've already passed that suggestion on to some other folks... I'll add this discussion to it...
		I've got to wrap up... much to do today.
09:02		Thank you all for coming, and I'll see you soon
	Tonya Souther	Enjoy yourself, Oz.
	Techwolf Lupindo	So, the proper way to go about it, ask if the object metadata can be used to add features down the road by LL.
	draconis.neurocam	take care oz
	oobscure	bye
		thanks oz