User:Andrew Linden/Office Hours/2010 07 13

Transcript

[10:59]	Ardy Lay:	Hi Simon
[10:59]	Simon Linden:	hello
[11:00]	Sindra Aquila:	hi simon
[11:00]	Welcome to	Linden office hours
[11:00]	Bronson Blackadder:	thank you meeter
[11:00]	Lady Sakai:	Hi Simon
[11:01]	Simon Linden:	I think Andrew is going to be a little late, he was getting called into another meeting
[11:01]	Rex Cronon:	hello everybody
[11:01]	Simon Linden:	it's about the 1.40 debugging
[11:01]	Rex Cronon:	u guys have so many meetings
[11:01]	Lady Sakai:	Hi Rex :)
[11:01]	Rex Cronon:	hi lady
[11:02]	Simon Linden:	people tend to schedule them around now since it's the middle of the day and there's good overlap with the Boston office
[11:02]	Youri Ashton:	hello
[11:02]	Youri Ashton:	oh great, simon is already here
[11:02]	Youri Ashton:	mind if i sended you a notecard simon?
[11:02]	Youri Ashton:	before the OH starts?
[11:02]	Rex Cronon:	the thing is. too many meetings can affect productivity
[11:02]	Youri Ashton:	its about a big problem i noticed earlier today
[11:03]	Youri Ashton:	Rex suggested to poke you about it
[11:03]	Youri Ashton:	i just sended Philip as well a copy
[11:03]	Youri Ashton:	got him in my friends list anyways =P
[11:03]	Rex Cronon:	simon and andrew r IMO the best qualiffied to find out what is wrong:)
[11:04]	Youri Ashton:	true
[11:04]	Youri Ashton:	still
[11:04]	Rex Cronon:	hi andrew
[11:04]	Youri Ashton:	little extra presure might be good ;)
[11:04]	Youri Ashton:	haha
[11:04]	Andrew Linden:	That's exactly what Simon and I want. Pressure from Philip.
[11:04]	Youri Ashton:	andrew, ill send you a copy as well
[11:04]	Andrew Linden:	Send him email, everyone!
[11:04]	Qie Niangao:	lol
[11:05]	Youri Ashton:	it isnt a small problem
[11:05]	Andrew Linden:	Just kidding. Actually he's making an concious effort to NOT lean on individual developers.OT
[11:05]	Youri Ashton:	so i didnt poke philip for no reason
[11:05]	Rex Cronon:	too much pressure not good man
[11:05]	Andrew Linden:	Ok some news...
[11:05]	Youri Ashton:	it took me about 3 hours to get out of grasmere... BEING 100% NUDE!!!!! ><
[11:06]	Andrew Linden:	I ran some tests last night and this morning on some combat regions running on the latest server-1.40
[11:06]	Andrew Linden:	and it looks like the performance blocker issue is fixed
[11:06]	Rex Cronon:	yeeeeaaaahhhhhh!!!!!!!!
[11:06]	Andrew Linden:	we're 95% sure. We're going to run one more test today on a troublesome region to see.
[11:06]	Techwolf Lupindo:	Yesterday I got stuck in a sim I could not tp out of. Ended up logging otu and logging into home position to get out of that sim.
[11:07]	Andrew Linden:	There is one "new" problem in 1.40 that showed up in the improved tests...
[11:07]	Youri Ashton:	tech, ill send a notecard, please read and tell me if you experienced the same or something similar
[11:07]	Rex Cronon:	u had similar problem to youris
[11:07]	Andrew Linden:	When the physics engine is under extreme load it has some "emergency rescue measures" that it uses to help recover.
[11:08]	Andrew Linden:	They are (1) Dropping the level of detail (LOD) of some objects
[11:08]	Andrew Linden:	(2) Putting some dynamic objects into "quarantine" where they are pinned in place temporarily
[11:08]	xstorm Radek:	odd
[11:09]	Andrew Linden:	and (3) (NEW!) delays adding new object's RigidBody's to the Physics World when in Quarantine mode
[11:09]	Sindra Aquila:	in relation to the tests is any research being done about the bounce problem and a solution for it?
[11:09]	xstorm Radek:	i just crashed and lost 500 lindens
[11:09]	Youri Ashton:	told ya, i aint the only one with those problems
[11:09]	Andrew Linden:	So the new behavior (3) appars to be breaking some content, but only when the physics engine is under load
[11:09]	Rex Cronon:	sorry to hear that xstrom. how did it happen?
[11:10]	Rex Cronon:	can instead make those objects phantom for a while?
[11:10]	Kitto Flora:	What sort of TD when the load is that high, Andrew?
[11:10]	Andrew Linden:	if scripts try to apply force on rez, or add some actions immediately after rez but before the rigidbody's have been added then some LSL calls can fail
[11:10]	Liisa Runo:	also (2) is causing problems sometimes
[11:10]	Sindra Aquila:	does this mean worse bouncing?
[11:11]	Andrew Linden:	Sindra, your "bounce problem" is the arrow that does not stop fast enough on impact?
[11:11]	Andrew Linden:	Sindra,. there is actually a feature in the works that will solve your problem, I think.
[11:11]	Rex Cronon:	how can a script know if "rigidbody's have been added"?
[11:11]	Sindra Aquila:	yes, the script os not setting it tp phys fast enough and the arrows go wild as if not scripted until; the script works
[11:12]	Andrew Linden:	Falcon is working on allowing you to set the "restitution" (bounciness) of object material to anything in the range of [0.0, 1.0], I think.
[11:12]	Liisa Runo:	neat
[11:12]	Youri Ashton:	hey open
[11:12]	Youri Ashton:	nice landing ;)
[11:12]	Youri Ashton:	haha
[11:12]	Opensource Obscure:	not? :D
[11:12]	Moon Metty:	:)
[11:12]	Andrew Linden:	So a restitution of 0.0 would help you out. Dunno when that feature idea would ship -- definitely not in 1.42.
[11:12]	Youri Ashton:	lol
[11:13]	Sindra Aquila:	ok, but the new 3 measures do not make it worse meanwhile?
[11:14]	Andrew Linden:	No Sindra I don't think the RCCS affects your bounciness problem.
[11:14]	Sindra Aquila:	thanks andrew
[11:14]	Andrew Linden:	Next bit of news...
[11:14]	xstorm Radek:	ok this is odd i can not tp out
[11:14]	Youri Ashton:	lol
[11:14]	Youri Ashton:	its the same i noticed earlier
[11:14]	Andrew Linden:	I'm going on vacation before the end of this week and I'll be out for about two weeks.
[11:14]	xstorm Radek:	and im getting a error
[11:14]	Youri Ashton:	can you cross the sim border?
[11:14]	Qie Niangao:	xstorm: crash yourself, maybe you'll make a profit this time.
[11:14]	Andrew Linden:	So I won't be at OH until... about July 30
[11:14]	Youri Ashton:	or even see sims nextdoor?
[11:14]	xstorm Radek:	Andrew you broke me yester day
[11:15]	xstorm Radek:	lol
[11:15]	Andrew Linden:	xstorm what broke?
[11:15]	xstorm Radek:	some please tell me what this is
[11:15]	xstorm Radek:	Uh oh. It's buggy. 500 Internal server error Sorry, something unexpected happened. Please try again in a few minutes.
[11:15]	Roach Gumbo:	leave a cell number so we can get in touch with you
[11:15]	Youri Ashton:	read my notecard andrew and you know
[11:15]	Youri Ashton:	i am not the only one with the problem, seems that more people have the same problem
[11:16]	Simon Linden:	What viewer are you using?
[11:16]	Youri Ashton:	and to prevend everyone having it...
[11:16]	Liisa Runo:	happy holiday :) if you go to some exotic place, bring me local stone
[11:16]	xstorm Radek:	the built in one
[11:16]	Youri Ashton:	at the time of the problem i was using 2.0
[11:16]	Youri Ashton:	currently using emerald
[11:16]	xstorm Radek:	Second Life 2.1.0 (206543) Jul 2 2010 13:54:58 (Second Life Beta Viewer)
[11:16]	Youri Ashton:	now i do NOT have a problem
[11:17]	Youri Ashton:	he is using the same that i was using
[11:17]	Cummere Mayo:	andrew while youre gone, can you remain logged in so we can mollest your avatar while youre gone?
[11:17]	Sindra Aquila:	neither had i with snowglobe
[11:17]	xstorm Radek:	i think i have a little problem
[11:17]	xstorm Radek:	i can not log out
[11:17]	Youri Ashton:	and no, that notecard is not a joke, unfortunately it is the truth
[11:18]	Simon Linden:	usually problems like the one you describe are due to network troubles or something with the back-end systems, not the regions themselves. It's more in how they are talking with other parts of the grid
[11:18]	Youri Ashton:	but for 3 hours long???
[11:18]	Simon Linden:	xstorm - do you get an error message?
[11:18]	Andrew Linden:	Yeah, Simon said what I was thinking.
[11:19]	Youri Ashton:	rather seem to be very odd
[11:19]	Youri Ashton:	think there is much more to it then just connection...
[11:19]	xstorm Radek:	it seem my viewer is working or some thing
[11:19]	Andrew Linden:	There is a "persistent presence" bug that causes a avatar to be "logged in" indefinitely until some attachments manage to save.
[11:19]	xstorm Radek:	im going to need to kill the task it seems
[11:19]	Youri Ashton:	everything seems to be okay, no lag after relog, sim fps was perfect, everything seems to be perfect
[11:20]	Youri Ashton:	and still it does what the notecards tells you
[11:20]	xstorm Radek:	oh great
[11:20]	Youri Ashton:	so connection.... i really have a hard time believeing that
[11:20]	Simon Linden:	So what happens when you try to log out? Error dialog? nothing?
[11:20]	xstorm Radek:	yes some thing is missing from the database it seems
[11:21]	xstorm Radek:	not a thing its stuck
[11:21]	Youri Ashton:	either normal log out, crashing log out, or it crashes before i even can relog
[11:22]	Youri Ashton:	it is a sirious problem which obvisouly needs more time to investigate
[11:22]	Andrew Linden:	xstorm can't logout because something is missing from database?
[11:23]	Andrew Linden:	Yes Youri, it sounds like a serious problem.
[11:23]	Second Life:	Select residents to share with.
[11:23]	Andrew Linden:	What was the date and hour that the problem happened, and what region name?
[11:23]	Youri Ashton:	as i been telling the AWgroupies during the group meeting
[11:23]	Second Life:	Select residents to share with.
[11:23]	Youri Ashton:	region name is Grasmere
[11:24]	Second Life:	Items successfully shared.
[11:24]	Andrew Linden:	oh right. and the hour?
[11:24]	Youri Ashton:	and time... euhm... about 8 hours ago now
[11:24]	Andrew Linden:	ok so... 03:00 this morning
[11:24]	Youri Ashton:	Pooley seemed to be a problem as well
[11:24]	Youri Ashton:	around that time..
[11:24]	Andrew Linden:	Ok. I don't have any new info or insights on that problem right now.
[11:25]	Andrew Linden:	I've got another item on my agenda today...
[11:25]	Andrew Linden:	SVC-93...
[11:25]	ROTATION and	llSetRot incorrectly implemented for child prims
[11:25]	Moon Metty:	o_O
[11:25]	Rex Cronon:	u fixed that:)
[11:25]	Andrew Linden:	I had implemented a partial fix for it and reviewed it with another LL dev
[11:25]	Andrew Linden:	and he had some questions...
[11:25]	Youri Ashton:	well, as you may have read, there is a ticket number attached. just in case, also ALL pictures included
[11:25]	Andrew Linden:	and I didn't know the answer, I was hoping you could help me out..
[11:26]	Andrew Linden:	first... how I "fixed" it...
[11:26]	Youri Ashton:	please do keep me up to date about that subject
[11:26]	xstorm Radek:	wow what a long list of errors that was
[11:26]	Andrew Linden:	All I did was add a PRIM_ROT parameter so that llSetPrimitiveParams() would work correctly on child prims
[11:26]	xstorm Radek:	some i never seen
[11:26]	Rex Cronon:	?
[11:27]	Andrew Linden:	so PRIM_ROT was a "working correctly" version of PRIM_ROTATION
[11:27]	Moon Metty:	yes
[11:27]	Liisa Runo:	i agree, that is the proper way to "fix it
[11:27]	Andrew Linden:	and the rotations that it returned and accepted were rotations in the region frame
[11:27]	Andrew Linden:	however, I was asked... maybe PRIM_ROT should apply in the object local frame rather than the world frame
[11:27]	Andrew Linden:	and I was wondering what you all thought of that
[11:27]	xstorm Radek:	test 11:28 AM PDT
[11:28]	Rex Cronon:	yes local
[11:28]	Moon Metty:	mmm, can we have both?
[11:28]	Andrew Linden:	that is, suppose I just add a PRIM_ROT that is object local but works correctly
[11:28]	Andrew Linden:	btw, I can't fix llSetRot() on child prims... have to wait for script versioning for that
[11:29]	Liisa Runo:	yea, keep the old school llSetRot as it is, just give us the new PRIM_ROT and we are happy and bring you candy
[11:29]	Kitto Flora:	What is 'set pos' on a child prim? Is it local or global position?
[11:29]	Liisa Runo:	local
[11:29]	Kitto Flora:	The 'set rot' should be local also
[11:30]	Andrew Linden:	You know, I had originally implemented both PRIM_ROT and PRIM_ROT_LOCAL in my "fix" but removed the local because I wanted to minimize my changes to improve the chance of them being accepted
[11:30]	office hours	is half over
[11:30]	Moon Metty:	yes ok
[11:30]	Andrew Linden:	So... PRIM_ROT on the root prim should be in region frame, and PRIM_ROT on child should be local frame?
[11:31]	Moon Metty:	the difference is only one rotation though (in the right direction)
[11:31]	Kitto Flora:	if the position stuff works correctly, then the rrotation stuff should match it, else it gets confusing
[11:31]	Moon Metty:	good point Kitto
[11:32]	Andrew Linden:	yes but... then the behavior depends very much on wether the prim is root or child...
[11:32]	Andrew Linden:	that is, the pos behavior is kinda confusing
[11:32]	Rex Cronon:	the root rotates in world coordinates.
[11:32]	Andrew Linden:	really we need PRIM_POS_REGION_FRAME, PRIM_POS_LOCAL_FRAME, etc
[11:32]	Liisa Runo:	yea, give them to us :)
[11:33]	Rex Cronon:	oh. so we can make prims rotate around root:)
[11:33]	Moon Metty:	maybe you can put the dilemma on the jira, so people can share their thoughts while you're on holiday
[11:34]	Andrew Linden:	btw, other LL devs object to stuff like PRIM_ROT_REGION and PRIM_ROT_LOCAL multiplicity because it expands the LSL API, where the right local rotation can be computed within the script so it would be possible to get by with just the region frame rotation param
[11:35]	Andrew Linden:	That is, why complicate the LSL API if there is a decent way to implement.
[11:35]	Kitto Flora:	Indeed. Only good argument would be if it were less load to do it 'inside LSL'
[11:36]	Andrew Linden:	Compared to Sebastean's ask for "uniform scaling" which can't really be done any other way effectively.
[11:36]	Rex Cronon:	u want us to use lsl to compute rotations? IMO that adds lag:(
[11:36]	Kitto Flora:	If theres no load advantage, why complicate LSL because some cant figute out 3-d geometry?
[11:36]	Rex Cronon:	uniform scaling?
[11:36]	Simon Linden:	scaling a full linked object in one operation
[11:36]	Rex Cronon:	oh
[11:37]	Andrew Linden:	Uniform scaling is what happens when you drag the little white corners of the UI scaling tool on a linked object.
[11:37]	Rex Cronon:	that is another thing that would reduce lag
[11:37]	Andrew Linden:	Right. I'm just saying that there really isn't a good way to do uniform scaling without a special LSL call
[11:37]	Kitto Flora:	Where cant a script scale a linset?
[11:38]	Andrew Linden:	whereas local-->region rotations can be computed
[11:38]	Andrew Linden:	these are arguments that other LL dev's use to shoot down or accept a proposed LSL API change
[11:39]	Andrew Linden:	Kitto, you can't reliably scale an entire linked object uniformly from a single script
[11:39]	Andrew Linden:	you need multiple scripts to do it, and it isn't reliable
[11:39]	Sindra Aquila:	but now the scalijng of say hair is done with a multitude of scripts in it, in every link does that not cause excessive lag, instead of one scaler?
[11:39]	Rex Cronon:	i can reliably scale an entire linked set uniformly from a single script:)
[11:39]	Kitto Flora:	Really? well - I have been scaling 3-pim linksets ok.
[11:40]	Andrew Linden:	Problems happen when you get near the linkability limit.
[11:40]	Kitto Flora:	OK, I nowhere near the limit
[11:40]	Rex Cronon:	that problem can be overcome:)
[11:40]	Andrew Linden:	It may be that you get half the scaling done and then it starts to fail
[11:40]	Moon Metty:	minimum prim-size is also a problem
[11:41]	Youri Ashton:	why problems near link limit?
[11:41]	Andrew Linden:	right... which reminds me the minimum scale size may have been the culprit in someone's linkability problem from last week
[11:41]	Liisa Runo:	/me voters for the simplified link limit, one stable number. Prim sizes not affecting the limit.
[11:41]	Youri Ashton:	i never have a problem there, unless it's too far away to link....
[11:41]	Andrew Linden:	someone was having trouble scaling something down small
[11:41]	Rex Cronon:	that was me
[11:41]	Rex Cronon:	and i sent u an example
[11:41]	Andrew Linden:	and it may have ben caused by one of the sub prims being too small to scall further down
[11:42]	Rex Cronon:	but i found an way around it:)
[11:42]	Rex Cronon:	no. it was the distance too root
[11:42]	Youri Ashton:	oh i have a feeling what the problem might be then
[11:42]	Rex Cronon:	plus small size of root and that link
[11:42]	Andrew Linden:	ok, I'm glad you figured it out Rex.
[11:43]	Andrew Linden:	btw, Simon and I have to leave at 11:45 on Tuesdays.
[11:43]	Rex Cronon:	:)
[11:43]	Youri Ashton:	when something gets to 0.010 meter, you can not size the rest down either
[11:43]	Sindra Aquila:	will not a uniform scaler prevent makers from needing resizers in almost anything from hair to whatever, in all prims making some ppl a walking script library on their own a uniform scaler might solve this
[11:43]	Andrew Linden:	I should re-iterate at the end of the meeting... I'll be on vacation for the next two weeks and won't be at office hours, but Simon will be here
[11:43]	Andrew Linden:	and maybe some other Lindens will show up while I'm gone.
[11:43]	Moon Metty:	have a great holiday Andrew :)
[11:44]	Sindra Aquila:	have a fine holliday simon
[11:44]	Rex Cronon:	hmm. i use only one script to resize the whole linked set, and do it fast. maybe u will find on sale soon, sindra:)
[11:44]	Andrew Linden:	Simon maybe we should recruit Kelly to come in for treatment.
[11:44]	Moon Metty:	lol
[11:44]	Youri Ashton:	have a great holiday simon and anderew!
[11:44]	Cummere Mayo:	have a great vacation andrew, bring us souvineerz
[11:44]	Rex Cronon:	that is until a uniform resize lsl function is available
[11:44]	Youri Ashton:	please do give the notecard to the one that may be ablet o solve it
[11:44]	Simon Linden:	I'll see if we can find some innocent Lindens to come and visit :)
[11:44]	Sindra Aquila:	yes, but lot bought stuff has resixers in about a hundred or whatever prims
[11:44]	Youri Ashton:	it can have a big impact on the community
[11:44]	Andrew Linden:	Simon, are you also going on vacation over the next two weeks?
[11:44]	Rex Cronon:	have a nice one, andrew
[11:44]	Opensource Obscure:	have fun Andrew. Thanks to you and Simon for the meeting
[11:45]	Simon Linden:	Thanks everyone for coming, see you next time
[11:45]	Qie Niangao:	Thanks Andrew and Simon.
[11:45]	Opensource Obscure:	bye bye everybody
[11:45]	Sindra Aquila:	thanks
[11:45]	Moon Metty:	see you soon :)
[11:45]	Moon Metty:	and thanks!
[11:45]	Simon Linden:	I should be around for the next 2 weeks, I think
[11:45]	Rex Cronon:	tc simon, andrew

Generated with SLog Wikifier