Bug triage/2007-07-30/Transcript

Agenda

Transcript of Rob Linden's office hours w/guest host Aric Linden:

[15:12]	Soft Linden:	Dzonatas - I don't know anything about a video
[15:12]	There is no suitable surface to sit on, try another spot.
[15:12]	Aric Linden:	Dzonatas, i don't know anything about the OSCON video, I'm sorry to say
[15:12]	Squirrel Wood:	VWR-Crash
[15:13]	McCabe Maxsted:	so where do we start?
[15:13]	Aric Linden:	heya bridie. did seattle just explode?
[15:13]	Dzonatas Sol:	aw... would be nice to see
[15:13]	Dzonatas Sol:	the vid
[15:13]	Bridie Linden nods
[15:13]	Bridie Linden:	The tubes are stuck up here
[15:13]	Squirrel Wood:	Ut-oh
[15:13]	Aric Linden:	Dzonatas, I suspect they'll show it eventually. Patience, my friend.
[15:13]	Wyn Galbraith:	Time for a new prim.
[15:14]	Daedalus Young:	PrimTime
[15:14]	Aric Linden:	Bridie, I think we should start. Can you make sure Rob is ok with that?
[15:14]	Aric Linden:	ahh. never mind
[15:14]	Bridie Linden:	lol
[15:14]	Able Whitman:	tubes: unstuck
[15:14]	McCabe Maxsted:	welcome back rob
[15:14]	Wyn Galbraith hopes we don't get swamped now.
[15:15]	Rob Linden:	let's try this again
[15:16]	Rob Linden:	so, let's go ahead and get through the misc pool, and then figure out what to cover next
[15:16]	Rob Linden:	VWR-289
[15:17]	Bridie Linden:	Apologies for the late posting of last weeks transcript and notes
[15:17]	Squirrel Wood:	mac related
[15:17]	Aric Linden:	Seems like a bug to me.
[15:17]	Aric Linden:	erm, it seems like something we should take on is what I should've said
[15:17]	Able Whitman:	if i understand 289 correctly, it looks like the viewer should probably just make sure that media urls are always urlencoded, right?
[15:18]	Dzonatas Sol:	is this the url from the media settings or as requested by the client side?
[15:18]	Rob Linden:	I just want to make sure I understand this....the complaint is that we're being more permissive than RFC1738 says we should be, right?
[15:18]	Wyn Galbraith waves to Khamon.
[15:18]	Squirrel Wood:	I read it as: URLs that do not follow rfc1738 standards don't work ?
[15:18]	Soft Linden:	The problem is it's different between Mac and Windows.
[15:19]	Soft Linden:	Windows accepts the wrong encoding, Mac does not.
[15:19]	Aric Linden:	um, i think the problem is that the mac and windows see spaces differently
[15:19]	Khamon Fate waves back
[15:19]	Aric Linden:	spaces in urls, that is
[15:19]	Aric Linden:	right soft?
[15:19]	Able Whitman:	well, windows also accepts the correct encoding, so alwaqys using urlencoding should be okay for both platforms
[15:19]	Squirrel Wood:	so the solution would be to change every space to %20 ?
[15:19]	Soft Linden:	I don't know about the specific bad part being accepted - that would make sense though
[15:19]	Soft Linden:	Sure, if it's just spaces
[15:19]	Dzonatas Sol:	this issue doesn't give me any detail to exactly where the problem with url's start
[15:20]	Daedalus Young:	yes, on entereing a url, let the viewer convert it to percent encoding
[15:20]	Aric Linden:	um, IE is more forgiving.
[15:20]	Rob Linden:	I think I get it: someone sets up something on windows, and it works, but they aren't following RFC 1738, so it fails on Mac
[15:20]	Able Whitman:	my guess is that it's not just limited to spaces; i'd prefer using a proper urlencode rather than special-casing spaces
[15:20]	Aric Linden:	yes rob.
[15:20]	Bridie Linden nods
[15:20]	Soft Linden nods to Able
[15:20]	Aric Linden:	agrees with able and soft
[15:21]	Aric Linden:	Able, would you like to update the bug before we import?
[15:21]	Wyn Galbraith:	Able Soft, sounds like a software company.
[15:21]	Able Whitman:	sure
[15:21]	Aric Linden:	thanks.
[15:21]	Able Whitman:	no worries, aric
[15:22]	Dzonatas Sol:	the rfc1738 is one issue, but fixing for the parcel media stream or on the client side are two other issues... so if you import this as-is it is like a meta issue
[15:22]	Rob Linden:	ok, I guess we should wait for Able to update the issue, and then import
[15:23]	Rob Linden:	sounds like there's a problem there to be solved, though there may be some debate about how to solve it
[15:23]	Rob Linden:	(but we don't have to have that debate now)
[15:23]	Able Whitman:	well, dzonatas makes a good point.. there's two ways to look at this issue: one, that the viewer doesn't ensure urlencoding, or that the sim doesn't ensure it always sends urlencoding.
[15:23]	Able Whitman:	i will note this in the bug as well
[15:23]	Aric Linden:	I think we can let the developers figure it out.
[15:23]	Squirrel Wood:	urlencode on demand ?
[15:24]	Rob Linden:	we may want to have a conversation on sldev about this one...this sort of thing can get ugly
[15:24]	Rob Linden:	next up: WEB-67
[15:25]	Squirrel Wood:	looks like its already been imported?
[15:25]	Rob Linden:	yup. next
[15:25]	Rob Linden:	VWR-298
[15:25]	Aric Linden:	298 has already been discussed too, i think.
[15:25]	Squirrel Wood:	yep
[15:25]	Squirrel Wood:	SL-49770
[15:25]	Rob Linden:	also imported
[15:25]	Aric Linden:	I believe it's being worked on internally already
[15:25]	Rob Linden:	next
[15:26]	Daedalus Young:	link to the discussion is in the comments
[15:26]	Rob Linden:	VWR-301
[15:27]	Aric Linden:	301?
[15:27]	Soft Linden:	I'm not sure what the correct behavior is here, really. It used to be that you couldn't set things >95% transparent.
[15:27]	Squirrel Wood:	Transparent objects - touch events can not be triggered
[15:28]	Daedalus Young:	no, but I think it also goes for a full alpha channel on an image
[15:28]	Squirrel Wood:	client side its 95%. but scripts can set 100%
[15:28]	Soft Linden:	So it used to be scripts setting things 100% transparent actually set 95% or whatever the cap was.
[15:28]	Rob Linden:	ok, the original reporter closed the issue, and then Celierra reopened it
[15:28]	Dzonatas Sol:	Soft, not from the UI, from LSL you can set 100%, and you can use 100% alpha textures
[15:28]	Daedalus Young:	you can have a fully transparent object without setting the transparency
[15:28]	Able Whitman:	rob, sorry for the delay, but i've update VWR-289.
[15:28]	Soft Linden:	Dzonatas - that LSL behavior changed at some point in the last year. It used to cap transparency there.
[15:29]	Gigs Taggart:	grr love that "crash 2 minutes after login bug"
[15:29]	Able Whitman:	updated*
[15:29]	Rob Linden:	great thanks Able (wasn't expecting realtime results, myself)
[15:29]	Dzonatas Sol:	I guess 301 is really a new feature request
[15:29]	Gigs Taggart:	it's been more than a year soft
[15:29]	Daedalus Young:	but I think there are cases where you want transparent objects to be touchable and cases where you don't want it
[15:30]	Soft Linden:	The big problem is that changing this behavior now might break things relying on it. Is there any major current content that's broken because of this and not yet fixed from the LSL change?
[15:30]	Benja Kepler:	scam objects in front of atms for example is where you DON'T want to be able to touch transparent
[15:30]	Soft Linden:	Right.
[15:30]	Rob Linden:	so, can we agree that the current behavior is not a bug?
[15:30]	Soft Linden:	Yeah.
[15:30]	Dzonatas Sol:	not a bug
[15:31]	Able Whitman:	yes, i agree, not a bug.
[15:31]	Rob Linden:	k...as a feature request with 1 vote, it can go way on the back burner
[15:31]	Bridie Linden:	I will resolve as 'not a bug'?
[15:31]	Rob Linden:	maybe just move it to "feature request"
[15:31]	Bridie Linden nods
[15:32]	Rob Linden:	next up: VWR-304
[15:32]	Aric Linden:	might this one simply be a lag prob?
[15:32]	Dzonatas Sol:	7 min vid
[15:33]	Able Whitman:	it would be nice to have more specific data; does the flashing only happen with a particular attachment? does it depent on attachment point? etc.
[15:33]	Gigs Taggart:	will attachments even show with that off?
[15:33]	Gigs Taggart:	in mouselook, many people don't even turn on "view avatar"
[15:33]	Soft Linden:	Anyone have a time index o.O
[15:33]	Dzonatas Sol:	the video will take time to find the problem being brought up
[15:33]	Soft Linden:	Ah, around 1:12
[15:34]	Soft Linden:	Worse around 1:26
[15:34]	Rob Linden:	does this seem like a common or severe problem?
[15:35]	Daedalus Young:	that looks like the camera position and the object's pposition are out of sync
[15:35]	Squirrel Wood:	Could it be a graphics driver bug thing?
[15:35]	Daedalus Young:	camera is sometimes already on the new position
[15:35]	Rob Linden:	seems very likely to get marked "someday maybe" if we import it
[15:35]	Gigs Taggart:	this is not severe rob
[15:35]	Aric Linden:	I agree.
[15:35]	Gigs Taggart:	like I said I doubt many even enable "show avatar in mouselook"
[15:35]	Squirrel Wood:	I have it enabled :p
[15:35]	Aric Linden:	but i'd be willing to have someone in QA look at it to make sure
[15:35]	Gigs Taggart:	only when using a vehicle like this will anyone notice usually :)
[15:36]	Aric Linden:	Rob, it can be assigned to me and I'll ask someone to look at it for grins
[15:36]	Soft Linden:	It sure looks like the dome sometimes sorts in front of all the jet control stuff when that's happening. If that's all alpha stuff, it could just be our old friend, the alpha sorting problem.
[15:36]	Rob Linden:	okee dokee. since we have such a small list, it might be good to talk about this particular case
[15:36]	Gigs Taggart:	could be
[15:37]	Dzonatas Sol:	It maybe just physics lag
[15:37]	Dzonatas Sol:	hard to say.
[15:37]	Rob Linden:	there's a lot of issues that are "someday maybe" if left purely to Linden Lab to fix
[15:38]	Soft Linden:	I say anyone who flies alpha-heavy stuff up to cloud level deserves what they get! :)
[15:38]	Rob Linden:	I'm wondering what the best way to mark these issues on jira.sl.com
[15:38]	Aric Linden:	/grins
[15:38]	Soft Linden:	Just really low priority, I think.
[15:38]	Gigs Taggart:	there is a meta issue for "bugs that would be good for open source devs to look at"
[15:38]	Gigs Taggart:	you could use that
[15:39]	Bridie Linden:	# Gigs?
[15:39]	Aric Linden:	well, personally, i think things should be prioritized. Not everything can be fixed or at least not everything can be fixed in the same order
[15:39]	Gigs Taggart:	Don't know offhand... hmm
[15:39]	Gigs Taggart:	searching...
[15:39]	Rob Linden:	ok...we'll bump down the priority on these
[15:39]	Aric Linden:	but looking at them is key. *knowing* what the issues are is important
[15:39]	Bridie Linden:	Make it small and import?
[15:40]	Aric Linden:	that allows us to priortize more effectively
[15:40]	Aric Linden:	you can assign it to me, Bridie. I'll have Dan look at it to see if he can ascertain what the issue is
[15:40]	Gigs Taggart:	VWR-1577
[15:40]	Gigs Taggart:	is the open source metabug for bugs that mostly nicholaz thinks are compact and good for OSS devs to look at
[15:41]	Rob Linden:	thx gigs
[15:41]	Rob Linden:	let's move on: VWR-310
[15:41]	Rob Linden fears the "myspace" reference in the title
[15:41]	Aric Linden:	I *think* this got looked at last week in another triage
[15:42]	Soft Linden:	Yikes
[15:42]	Daedalus Young:	I can't remember it from another triage
[15:42]	Gigs Taggart:	aric all these are leftover bugs from an old triage
[15:42]	Able Whitman:	the right solution to 310 is to go back in time and prevent browsers from ever being able to auto-play background music, i think.
[15:42]	Gigs Taggart:	that were not addressed
[15:42]	Daedalus Young:	I know YouTube vids in profiles do stop after closing the profile
[15:42]	Soft Linden:	Well, it's linked to MISC-193 which isn't imported.
[15:42]	Soft Linden:	linked to/dupe of MISC-193
[15:42]	Daedalus Young:	and I know myspace has been buggy in the past
[15:42]	Aric Linden:	/could easily be misremembering
[15:42]	Daedalus Young:	if not it still is
[15:42]	Gigs Taggart:	iirc myspace uses flash to play music
[15:43]	Daedalus Young:	correct
[15:43]	Squirrel Wood:	Mayhaps that one could be fixed by loading "about:blank" on closing of the profile ?
[15:43]	Gigs Taggart:	might be the more general case that "flash movies keep playing"
[15:43]	Soft Linden:	Myspace actually lets them embed just about anything. I've seen WMP, etc used directly in there
[15:43]	Gigs Taggart:	or the sound of them
[15:43]	Daedalus Young:	well, youtube closes on exit, that's a flash
[15:43]	Gigs Taggart:	ok
[15:44]	Able Whitman:	maybe it is an fmod issue and not a browser issue?
[15:44]	Daedalus Young:	or at least that worked when I tested it
[15:44]	Gigs Taggart:	does mozlib use fmod sound or mozilla's own sound handling?
[15:44]	Dzonatas Sol:	310, import to ask llmozlib dev for a fix?
[15:44]	Soft Linden:	If we can ask for an URL, we can skip a lot of guessing. Needs more info, unless someone wants to experiment with pages?
[15:45]	Gigs Taggart:	yeah lets just resolve needs info and IM lex neva about it since he seems to have reproed it
[15:45]	Able Whitman:	well, the bug lists one user whose profile url seems to exhibit the problem
[15:45]	Soft Linden:	Missed that - cool!
[15:45]	Able Whitman:	although it doesn't list the speciffic url
[15:45]	Rob Linden agrees with Able
[15:45]	Rob Linden:	the repro is to visit a profile page, not a specific URL
[15:45]	Daedalus Young:	I can test it too, environment is Mac, that could also be a problem
[15:45]	Rob Linden:	(assuming, of course, the profile is the smae)
[15:46]	Gigs Taggart:	yeah we need to determine if it's platform specific
[15:46]	Able Whitman:	ah, jerome wishbringer has deleted their profile url :/
[15:46]	Able Whitman:	so we do need more info here.
[15:46]	Gigs Taggart:	I already IM lex offline
[15:46]	Gigs Taggart:	asked him to bump the bug with more info
[15:46]	Rob Linden:	ok...resolve/need info
[15:47]	Rob Linden:	next up....
[15:47]	Daedalus Young:	Doubledown Tandino has a myspace in his profile
[15:47]	Rob Linden:	SVC-72
[15:47]	Gigs Taggart:	Rob SVC-72 segue's nicely into what I wanted to talk about
[15:47]	Gigs Taggart:	if you want to just turn me loose there :)
[15:48]	Dzonatas Sol:	May is the last comment on 72
[15:48]	McCabe Maxsted:	I tried to repro it with my own profile nad couldn't
[15:48]	Rob Linden:	ok, go for it gigs
[15:48]	Gigs Taggart:	ok everyone I'd like to officially announce the "BLT" protocols/project.
[15:49]	Benja Kepler:	bacon...lettuce....
[15:49]	Gigs Taggart:	Bugs like SVC-72 seem very network related, but are hard to repro since our connections don't suck in general
[15:49]	Gigs Taggart:	so the idea behind BLT is to provide network emulation test scenarios for the reproduction of bugs related to packet loss, latency, low bandwidth
[15:49]	Rob Linden:	(my connection sucks, but I digress)
[15:49]	Gigs Taggart:	bugs like SVC-72 should be related to the BLT Meta-bug which is http://jira.secondlife.com/browse/MISC-506
[15:50]	Gigs Taggart:	the BLT Wiki page is https://wiki.secondlife.com/wiki/BLT
[15:50]	Gigs Taggart:	the idea is that we will develop network emulation scenarios, numbered and tracked, and use those to communicate the specific reproduction conditions, and the percentages occurance that the bug occurs at each relevant scenario
[15:51]	Rob Linden:	gigs, this looks great. have you successfully reproed something with this type of setup?
[15:51]	Gigs Taggart:	the network emulation is done with standard linux kernel tools, the TC and traffic shaping modules
[15:51]	Gigs Taggart:	Yes one thing Rob, the "scripts loading twice" showed up reliably at 200ms additional latency injected into the connection.
[15:51]	Able Whitman:	gigs, are there similar tools for windows which allow for the same sort of testing?
[15:51]	Gigs Taggart:	I believe a lot more bugs will be reliably reproduced
[15:52]	Gigs Taggart:	Able, probably. Network emulation is a complex thing though, so we need to make sure the tools can work exactly the same.
[15:52]	Able Whitman nods
[15:52]	Gigs Taggart:	ideally if you possibly can, you can use a linux box as an "ethernet bridge" with two interfaces
[15:52]	Gigs Taggart:	and then you can still develop on windows while using the same lag emulation tools
[15:53]	Gigs Taggart:	so that all packets from your windows box pass through the linux network lagger ethernet bridge box :)
[15:53]	Gigs Taggart:	http://linux-net.osdl.org/index.php/Netem
[15:53]	Aric Linden:	Gigs, this is really cool.
[15:54]	Aric Linden:	If you're interested, i'd be happy to work with you or have someone from my team spend some time with you on this.
[15:54]	Gigs Taggart:	thanks Aric, I hope we can get SL working better for international users who push 300+ms pings sometimes and have a lot of trouble.
[15:54]	Rob Linden:	see the links at the bottom of http://en.wikipedia.org/wiki/Traffic_shaping
[15:54]	Able Whitman:	right, that makes sense. unfortunately i don't have a spare machine but it's a good technique for those that do. regardless, this is a great plan and if i can participate i'd really like to.
[15:54]	Aric Linden:	we have an internal load testing initiative going on. this might make an interesting corresponding project.
[15:54]	Rob Linden wonders if there's a vmware image of dummynet
[15:55]	Gigs Taggart:	Able well you are free to use windows tools, it would be best to have someone confirm your repros using linux-netem if possible though, in case there are differences... once you get confident you have very similar setups though, you could put them right alongside the netem ones
[15:55]	Squirrel Wood:	heh. I'd like to see the sl client actually make use of my 16mbit connection for once ^^
[15:55]	Rob Linden:	with respect to SVC-72, what should be the resolution?
[15:56]	Gigs Taggart:	Rob I would relate it to MISC-506 and import probably :)
[15:56]	Soft Linden:	Aric - it's possible that this could be set up internally so that some VPNs ran through this. Then you could have any tester or developer take on an intentionally bad connection.
[15:56]	Gigs Taggart:	I have seen very similar reports
[15:56]	Aric Linden:	/nods at soft
[15:57]	Bridie Linden:	I linked as 'relates to' MISC-506
[15:57]	Gigs Taggart:	soft that would be cool, but there may be dozens of test scenarios, if you could do that with VPNs then that would work fine
[15:57]	Rob Linden:	Aric....we should also talk about how I've seen this done in a past life
[15:57]	Gigs Taggart:	there should be a "big 5" or so, say 300ms latency, etc
[15:57]	Aric Linden:	ok
[15:57]	Aric Linden:	erm, ok rob
[15:57]	Aric Linden:	I've also done something related in past lives
[15:57]	Gigs Taggart:	alright I'm done if anyone else has questions :)
[15:58]	Aric Linden:	gigs, let's chat offline sometime
[15:58]	Gigs Taggart:	k, I'm on efnet IRC, Gigs
[15:58]	Aric Linden:	you too rob.
[15:58]	Rob Linden:	re: SVC-72...any Lindens think this is in a state to import yet?
[15:58]	Aric Linden:	not I
[15:58]	Bridie Linden:	I'm not so sure...
[15:58]	Rob Linden:	k... I'd like to resolve/need info
[15:59]	Rob Linden:	one last note before going
[15:59]	Rob Linden:	Aric has graciously (foolishly?) agreed to take over these triage meetings from me
[15:59]	Gigs Taggart:	hehe
[15:59]	Wyn Galbraith cheers.
[15:59]	Aric Linden:	he made me do it
[15:59]	Phli Foxchase:	^^
[15:59]	Able Whitman:	lol
[16:00]	Soft Linden:	SVC-72 - I'm seeing a lot of Google hits on problems with UDP and that router. Looks worth asking Netgear about directly. They might be able to say exactly what the problem is.
[16:00]	Able Whitman:	does that mean Aric inhereits the plywood box of triage?
[16:00]	Wyn Galbraith:	Is Rob the devil then?
[16:00]	Gigs Taggart:	heh the triage box
[16:00]	Dzonatas Sol:	we look forward to you with future meetings =)
[16:00]	Aric Linden:	thanks.
[16:00]	Aric Linden:	I don't anticipate much changing.
[16:00]	Aric Linden:	Bridie will be helping out
[16:01]	Rob Linden:	I might start having a triage to specifically deal with jira/wiki/source code issues, but I should probably just get in the habit of dispatching those first, and then have a triiage as necessary
[16:01]	Dzonatas Sol:	A note about 72, sometime the MTU going through routers and NATs really cause havoc
[16:01]	Bridie Linden:	I will?
[16:01]	Aric Linden:	and I'm sure Soft will continue to bring his sagacity to us
[16:01]	Bridie Linden:	j/k
[16:01]	Soft Linden thumbsup!
[16:01]	Aric Linden:	thanks!
[16:01]	Bridie Linden:	aye, aye captain!
[16:01]	Aric Linden:	lindens never take anything very seriously
[16:01]	Rob Linden:	since we've drastically increased the number of triage meetings, I'll put off adding one just now, and prehaps work in a supporting role on the others
[16:02]	Rob Linden:	anyway, thanks everyone for coming!
[16:02]	Daedalus Young:	except when it comes to wagering ;)
[16:02]	Aric Linden:	Triage is the new bob dylan at linden
[16:02]	Rob Linden:	and thanks Aric, for taking over!
[16:02]	Wyn Galbraith:	No wagering, banned.
[16:02]	Daedalus Young:	thanks you too guys
[16:02]	Able Whitman:	thanks for an excellent triage, everyone!
[16:02]	Wyn Galbraith:	Thanks for having these meetings. :)
[16:02]	Aric Linden:	sure. i think, assuming it's ok with rob, that we'll continue to meet here for the time being.