Open Source Meeting/2009-04-23

• [14:00] Rob Linden: Hi folks!
• [14:00] Aimee Trescothick: hey
• [14:00] Thickbrick Sleaford: hi
• [14:00] Aimee Trescothick: hi Rob
• [14:00] Otaku Thor: hi Rob
• [14:00] Robin Cornelius: Hi everyone
• [14:00] Aimee Trescothick: was beginning to think there'd been a mass kidnapping via IRC or something :D
• [14:00] Aimee Trescothick: hi Robin
• [14:01] Otaku Thor: bad botnets .. lol
• [14:01] Rob Linden: no new agenda items this week, but there's plenty for us to talk about
• [14:02] Rob Linden: maybe we can bash out a quick agenda now:
• [14:02] Rob Linden: 1. Committer list - Mm was asking in the QA process JIRA about that
• [14:03] Rob Linden: 2. QA process. in particular, some feedback on how JIRA workflow should work (more on this in a bit)
• [14:03] Rob Linden: 3. Update on how the 1.23 merge is going
• [14:04] Rob Linden: that list would keep us busy for a little while
• [14:04] Rob Linden: oh....also:
• [14:04] Rob Linden: 4. easybuild dev plan
• [14:04] Rob Linden: lots of stuff to talk about there
• [14:04] Rob Linden: anyone have anything to add/substitute?
• [14:05] Opensource Obscure: fine : )
• [14:05] Rob Linden: (while I'm waiting for that: std disclaimer: everything you type in chat for this meeting will probably make its way onto wiki.secondlife.com, so mind that as you type)
• [14:06] Rob Linden: cool....since people aren't feeling chatty about the agenda itself, let's go down the list
• [14:06] Rob Linden: 1. Committer list
• [14:06] Rob Linden: there's a few folks that have commit access for legacy reasons and weird one-off reasons, so I wouldn't want to list their names as "committers"
• [14:07] Rob Linden: I guess I'm getting ahead of myself though....first some background
• [14:07] Rob Linden: looks up the jira task with Mm's comment
• [14:08] Rob Linden: see the first comment here: https://jira.secondlife.com/browse/VWR-12931
• [14:08] Rob Linden: Mm writes: "I've never been a big fan of QA through final testing. We had a discussion when Philip presented about open code reviews, but that got reduced to "a comment from another committer (not necessarily a Linden) who has thoroughly reviewed the patch, and is willing to stake their credibility as a committer that the patch should be safe for checkin." I don't think that qualifies as a code review. I don't even know who the approved committers are. Is there a list somewhere?
• In addition to quality assurance, the code review could also act as the gatekeeper. Right now, the criterion is: "Specific feature/patch proposals should be discussed on sldev@ (let at least one full business day pass without comment before considering silence assent)." It doesn't say what happens when someone says yes and someone else says no.
• So my suggestion is that QA going in will make QA going out a whole lot easier: open code reviews.
• Mike
• [14:09] Rob Linden: so, back to what I was going to say on the topic:
• [14:10] Rob Linden: I think he's right that we should have a list of people who are approved reviewers
• [14:10] Rob Linden: ...but I don't want to publish a "pester me" list for the people who already have commit access, because that wasn't part of the deal when they signed up
• [14:11] Robin Cornelius: everyone pesters me anyway, so i'm fine with that
• [14:11] Rob Linden: ....so, I'm thinking we should just have an opt-in list on the wiki of reviewers
• [14:11] Robin Cornelius: sounds fine
• [14:12] Youri Ashton: thank you for the slurl again soft!
• [14:12] Q Linden: maybe coupled with a call for folks to sign up
• [14:12] Q Linden: so that we have "enough" to start
• [14:12] Rob Linden: k....cool. who here has access that wants to be on that list?
• [14:12] Soft Linden: Youri: Sure. Where on the wiki are you seeing the old location?
• [14:13] Robin Cornelius: Rob, sure add me
• [14:13] Youri Ashton: i sended the link to your im
• [14:14] Rob Linden: Youri: is there someplace where this is still advtised as happening at my old inworld office?
• [14:14] Aimee Trescothick: I can't commit to it with me existing workload right now, but possibly in the future
• [14:14] Aimee Trescothick: me=my
• [14:15] Dirk Talamasca: Hiya Q!
• [14:15] Rob Linden: of the Lindens, I'll add Soft, Tofu, Q, Philip, Merov and myself
• [14:15] Q Linden: hey dirk
• [14:15] Robin Cornelius: Well just becuase my name is there does not mean i will be open to every asking, this could/should work like the debian developers/mentors ssyetm
• [14:15] Robin Cornelius: *system
• [14:15] Ellla McMahon: Hello everyone :))
• [14:15] Rob Linden: (unless one of them doesn't want to be added)
• [14:16] Q Linden: 'scool.
• [14:16] Soft Linden: All good!
• [14:17] Rob Linden: we'll say: "this isn't an exhaustive list, but is a good starting poing"
• [14:17] Youri Ashton: sorry rob, didnt notice the question yet. the link was actually on SL wiki
• [14:17] Youri Ashton: ill send the link in main this time
• [14:17] Youri Ashton: [1]
• [14:17] Youri Ashton: it might be handy to check all slurls
• [14:18] Youri Ashton: i did notice some changes in locations without the slurl being changed
• [14:18] Rob Linden: and I agree with Robin: a mentor system sounds about right, though I'll need to study up on how Debian works to know for sure if that's the right model to emulate
• [14:18] Rob Linden: thanks for the heads up Youri
• [14:18] Youri Ashton: no problem!
• [14:19] Robin Cornelius: basicly DDs have commit access to the ftp-masters others need a DD to get there work pushed to the servers, so a DD has to review a non DD's work,
• [14:19] Robin Cornelius: thats all i was refering too, probaly don't need any more depth than that
• [14:19] Arawn Spitteler: DD?
• [14:19] Youri Ashton: I am trying to get a job at LL myself, but unfortnuantly due somthing important in RL i didnt have the time today to get to that yet
• [14:19] Robin Cornelius: except we also add code revew between developers
• [14:20] Rob Linden: cool, ok....I'll put the committer list here when I get a chance: https://wiki.secondlife.com/wiki/HTTP_Texture_Development
• [14:20] Robin Cornelius: in fact this might even be closer to the linux kernel commit system
• [14:20] Thickbrick Sleaford: Arawn, DD is Debian Developer
• [14:21] Rob Linden: yeah, that sounds like a good model...we're modeling other aspects of this around Linux
• [14:21] Robin Cornelius: this is when i like git ;-) and its ability to email and sign off patches you approve
• [14:21] Aimee Trescothick: yes :)
• [14:21] Rob Linden: git has that built in? interesting
• [14:21] Arawn Spitteler: I suppose, actually, that's already done, with Linden Clients, where thrid party clients get vetted for their patches.
• [14:22] Arawn Spitteler: hopes his inverted parasyuntax is comprehent
• [14:22] Robin Cornelius: Rob its actualy helper scripts in theory Hg could be made to do the same
• [14:22] Youri Ashton: \
• [14:22] Youri Ashton: hi merov
• [14:23] Rob Linden: well, anyway, moving on to the QA process
• [14:23] Rob Linden: (item #2)
• [14:23] Rob Linden: Robin: great email to sldev@
• [14:23] Robin Cornelius: sorry BRB, one moment this font is making me feel sick
• [14:24] Rob Linden:  :)
• [14:24] Youri Ashton: Q: I have a question for you. Nix Linden told me you were working on something new called 'flexie scultie', could you tell me more about what the idea is and how far you are at this point?
• [14:24] Q Linden: that's not me, that's Qarl
• [14:24] Youri Ashton: ah, might be a simple mistake in identity :p
• [14:25] Q Linden: I believe the idea is essentially animated sculpties
• [14:25] Youri Ashton: sorry for that!
• [14:25] Q Linden: but supported
• [14:25] Rob Linden: to bring everyone up to speed: I sent mail a while back about QA process...specificly, when do we know its done?
• [14:25] Arawn Spitteler: Done?
• [14:25] Aimee Trescothick: pokes a ork and sees if the juices run clear
• [14:25] Aimee Trescothick: a fork works better though
• [14:26] Aimee Trescothick: orcs generally don't appreciate being poked anywhere
• [14:26] Robin Cornelius: Rob in my reply one of the things i was thinking about is this [2]
• [14:26] Arawn Spitteler: doesn't want any Orks salivating on his table
• [14:26] Thickbrick Sleaford: a very long stick
• [14:26] Rob Linden: I suspect our process won't be much more formal than orc poking
• [14:27] Robin Cornelius: we can at least count RC and major bugs and try to get them to 0
• [14:27] Youri Ashton: I come from the orcs. We eat with spoons and forks, we love to eat our pork!
• [14:27] Youri Ashton:  :)
• [14:27] Rob Linden: yeah, I think that's about right.
• [14:27] Rob Linden: heh
• [14:27] Youri Ashton: a little fun could be nice once ina while :)
• [14:28] Rob Linden: indeed
• [14:28] Robin Cornelius: there are feature targets that need to be met so thats easy to see if its achieved or not, then the rest is crash rate and bug count
• [14:28] Arawn Spitteler: Gamma Testing is customer Feedback, but QA should make it to a successful product run
• [14:28] Rob Linden: anyway....the triage process can be an optional thing (per Robin's point about volunteer time) and done somewhat asyncrhonously, but I think that's going to be part of it
• [14:29] Merov Linden: do we want "feature driven releases" or "time driven releases"?
• [14:29] Youri Ashton: might it not be handy to intergrate that in the SL viewer? Like under Help menu and then popping up a menu? It sure would get alot more attention
• [14:29] Rob Linden: should we try to develop test plans and rally people to run through them, or is that the kind of thing that you pretty much have to pay people to do?
• [14:30] Opensource Obscure: i'd like to have testing locations
• [14:30] Youri Ashton: now people wont go look for it since they have to do more then they like, people have the urge to be very lazy
• [14:30] Opensource Obscure: if that's what you mean Rob
• [14:30] Q Linden: It wouldn't hurt to put a test plan in the JIRA
• [14:30] Robin Cornelius: Merov thats a interesting question, i'm sure there are feature targets for a given release, but we also want a time frame
• [14:30] Q Linden: people could at least say "hey, did you think of THIS case?"
• [14:30] Q Linden: even if they don't run the test plan themselves
• [14:31] Opensource Obscure: reads [3]
• [14:31] Q Linden: OO, what do you mean by "testing locations"?
• [14:31] Robin Cornelius: plug in, switch on, check for fire and smoke
• [14:31] Youri Ashton: indeed, when its in the SL viewer (auto updated) then people can react and give their input for anything asked. This case LL will gain more answers most likely
• [14:31] Rob Linden: the toughest problem that I think we're going to have is *knowing* that the product is good enough at the end of the time we allot for ourselves
• [14:32] Youri Ashton: perhaps even more possibilities to solve things
• [14:32] Youri Ashton: it's just a thought tho
• [14:32] Rob Linden: ...it may be we just have to risk putting it out there, and then get embarressed every so often
• [14:32] Robin Cornelius: need to be able to bang out a quick "doh!" bug fix in those situations
• [14:33] Robin Cornelius: like most open source projects from time to rtime
• [14:33] Opensource Obscure: (Q: nevermind, i'll ask that later)
• [14:33] Rob Linden: this is one of the reasons why companies hire QA departments, so they know more about the quality of the product before putting it out to a wide audience, but this project is going to be about experimenting with newer community-driven process
• [14:33] Rob Linden: Robin: not exclusive to open source projects, as it turns out ;-)
• [14:33] Robin Cornelius: no but a more transparent process
• [14:33] Rob Linden: Quality is Job 1.1
• [14:34] Youri Ashton: it might be a risk Rob, but it better be a risk to be embarressed then to wait alot time for an answer you need asap
• [14:35] Rob Linden: ok....so, I think we'll go with a bit of an ad hoc process. we'll make sure we've at least looked through all of the bugs marked "Affects version: http-texture branch" in JIRA, and decided to either fix or live with all of them
• [14:35] Rob Linden: we'll look at the crash rates, we'll get the crash stacks posted in JIRA
• [14:36] Rob Linden: we'll declar a "pencils down" date where, unless there's a known showstopper, we press forward
• [14:36] Youri Ashton: i am not vermilliar with the subject
• [14:37] Rob Linden: ...and we'll be prepared to issue a quick followup release if we have a brown paper bag bug
• [14:37] Rob Linden: [(http://catb.org/~esr/jargon/html/B/brown-paper-bag-bug.html] )
• [14:37] Youri Ashton: ty
• [14:38] Youri Ashton: ah, lol
• [14:38] Rob Linden: ok...here's the one other big QA issue that I'm trying to wrap my head around....
• [14:38] Rob Linden: marking issues as "Resolved/Fixed" in this build
• [14:38] Thickbrick Sleaford: (that definition sounds like the debian ssh fiasco)
• [14:38] Youri Ashton: what is the issue to be precise?
• [14:39] Rob Linden: that will inevitiably confuse people who want to see them fixed in the main Second Life viewer and don't care about this viewer
• [14:39] Arawn Spitteler: REsolved should carry a comment, i any Jira
• [14:40] Robin Cornelius: fixed internally causes enough confusion
• [14:40] Arawn Spitteler: Treage resolves in ways that don't make sense, to someone who hasn't attended.
• [14:40] Youri Ashton: still dont get closer to the problem :p
• [14:40] Q Linden: yah...I think we're going to need another state or another field
• [14:40] Opensource Obscure: creating a second similar ticket with 1.23 only in Version and closing it while leaving the other one open?
• [14:40] Q Linden: either "resolved in open source viewer" or a second resolved field
• [14:41] Rob Linden: I'm coming to the conclusion that we'll probably be best off with a spearate project
• [14:41] Robin Cornelius: that could worl
• [14:41] Rob Linden: ...and then copy/link issues that are also problems in the main Second LIfe viewer
• [14:41] Robin Cornelius: work
• [14:41] Youri Ashton: everything can be solved Rob :)
• [14:41] Youri Ashton: it is just how you put your mind to things :)
• [14:42] Robin Cornelius: i like that idea more and more, helps keep the name space clean, easy to filter and find issues
• [14:42] Youri Ashton: you can think of a difficult solution but in most cases a simple solution always works better
• [14:42] Rob Linden: it pains me a little to offer that solution, because the viewer is all one big happy codebase, but I don't see a way around
• [14:42] Thickbrick Sleaford: and resolved is kinda boolean in nature
• [14:42] Robin Cornelius: its in the same JIRA so its all linkable
• [14:42] Aimee Trescothick: yeah, I think it's they only way of avoiding greater pain and bloodloss though
• [14:43] Aimee Trescothick: -y
• [14:43] Robin Cornelius: it might be worth having a field to specificy track 1-1 related issues rather that the usual, related to by etc
• [14:43] Rob Linden: yeah, I think so. that's probably where we'll end up, but I'd like to kick the idea around a little more
• [14:43] Rob Linden: plus we need to announce the name we've picked out for the project before it'll make sense to create the new project
• [14:44] Rob Linden: so, in the interim, we'll just need to use "http-texture"
• [14:44] Rob Linden: which will be the "affects version", and leave it in the same project temporarily
• [14:44] Merov Linden: 1 question Rob
• [14:44] Robin Cornelius: whens the name being announced
• [14:45] Merov Linden: if we create a new project, we can still link bugs from different projects together right?
• [14:45] Youri Ashton: if i may offer an idea, wouldn't it be better to check everything, and i do mean everything, and clear out the stuff that is not needed anymore, make a CLEAN viewer that doesnt have junk in so to speak. and from there test it, if it works, then it should be fine, if not, then try to fix each bug on its own, perhaps some bugs are related and can be fixed by simple adaptions. it might be a good solution, although it will take a lot of time to do this
• [14:45] Rob Linden: Robin: soon I hope...there's some i dotting and t crossing that has to happen first, and I need to check status on that
• [14:45] Harleen Gretzky: Why not OSS- ?
• [14:46] Youri Ashton: i used to make modifications for games like Red Alert 3, C&C generals, and several others, i always checked the coding first for any possible problems
• [14:46] Robin Cornelius: meh, its a lot harder to get some one to blurt out something my mistake in text chat, as you would in voice ;-p
• [14:46] Rob Linden: heh
• [14:46] Robin Cornelius: Yorui, have you seen how much code is in the viewer?
• [14:46] Youri Ashton: yes i seen :)
• [14:46] Robin Cornelius: + its interlinked criticaly with the server via the messages which may not be bug free either
• [14:46] Youri Ashton: as i said, it might be alot of time, but it might be worth it aswell
• [14:46] Rob Linden: Youri: [4] <- my standard answer
• [14:48] Q Linden: lol, rob, I was gonna post that link
• [14:48] Rob Linden:  :)
• [14:48] Rob Linden: I know exactly how to search for it these days
• [14:48] Youri Ashton: i didnt say from scratch, i do however say that the SL viewer needs a good check up, perhaps someone finds a little thing that might be the problem, a wrong letter, a wrong sighn, anything. it could be simple. but unfortunatly it takes a lot of time
• [14:48] Rob Linden: I cite it a lot
• [14:49] Youri Ashton: it was just an idea anywayz :p
• [14:49] Robin Cornelius: but out of the ashes came what we now know as firefox
• [14:49] Robin Cornelius: is being difficult
• [14:49] Rob Linden: anyhoo....Merov's question about linking: yes we can link together issues in different projects
• [14:49] Rob Linden: so, that shouldn't be a problem
• [14:50] Merov Linden: wasn't sure about that
• [14:50] Merov Linden: so, I'm +1 on creating a different project
• [14:50] Youri Ashton: i just like to add as final thing, as I can do it with the coding of several games on my own and find problems, then it shouldnt be a problem for proffesionals ;)
• [14:50] Rob Linden: Robin: sadly, the company formerly known as Netscape can hardly be found in the new project
• [14:50] Latif Khalifa: some long term re-architecture of the viewer would not be a bad idea... like abstracting the protocol handling (communiction with the server) which would open up for some exciting additions such as a plugin architecture
• [14:51] Youri Ashton: just to be sure, i dont mean it wrong in any way :)
• [14:51] Rob Linden: Latif: you've hit on Q's favorite topic I think :)
• [14:51] Arawn Spitteler: There's a lot wrong, in SL; Known issue
• [14:51] Q Linden: latif, that's pretty much what we've been doing this year (see notifications) and what several people are working on Even As We Speak
• [14:51] Merov Linden: on rearchitecture, the thinking is to add unit test which, usually, involves a good deal of refactoring and clean up
• [14:52] Latif Khalifa: Q great news :)
• [14:52] Rob Linden: we can go on a long time about the architecture thing, but before we do...
• [14:52] Merov Linden: so you get a cleaned up, tested and continuously testable class
• [14:52] Rob Linden: Merov: I hope we're not keeping you from doing the 1.23 merg
• [14:52] Rob Linden: merge that is
• [14:52] Q Linden:  :)
• [14:53] Robin Cornelius: that reads *really* badly
• [14:53] Youri Ashton: i personnaly think it might be a good idea to put a small team on this and just try it out, dont tell the public about it besides in meetigns like this for example. but do try, it might be a solution for some problems if not many
• [14:53] Merov Linden: I'll bail out in a sec to meet with CG :)
• [14:53] Youri Ashton: besides , everything can be redesigned and improved aswell
• [14:53] Latif Khalifa: CG needs to push those svn updates :D
• [14:53] Youri Ashton: oki merov, have fun!
• [14:53] Rob Linden: Cool, Merov!
• [14:53] Youri Ashton: CG is usually meditating at his place :p
• [14:54] Youri Ashton: i see him often there :p
• [14:54] Rob Linden: (and if that looked like whip cracking, trust me it wasn't....I'm sure Merov is as eager to get to that as anyone)
• [14:54] Youri Ashton: great guy tho :)
• [14:54] Latif Khalifa: i raally wish public svn was synchronized in some orderly fashion
• [14:55] Merov Linden: going to his next meeting...
• [14:56] Rob Linden: public svn: yeah, it's not as clean as we would like
• [14:56] Latif Khalifa: i am pestering soft all the time on #opensl
• [14:56] Rob Linden: it's going to be messy for a while because of a number of projects we're cooking up
• [14:56] Latif Khalifa: but i don't see a reason why we cannot have viewr_1.23 branch
• [14:56] Latif Khalifa: once BSI nightlies start coming out twice a week
• [14:56] Rob Linden: but I think when the dust settles, we'll be in a bettre place in that regard
• [14:56] Rob Linden: Latif: what's wrong with the 1.23 pushes?
• [14:57] Latif Khalifa: there are none
• [14:57] Latif Khalifa: i guess that's what's wrong with then ;P
• [14:57] Robin Cornelius: we have the branch but its not updated
• [14:57] Rob Linden: hrm....ok, I guess that was underway. that wasn't by design that it's not working
• [14:57] Soft Linden: Pinged CG a couple times - but really the surest way to see it done is to file a JIRA and assign it to him
• [14:58] Latif Khalifa: well it happens every RC cycle
• [14:58] Robin Cornelius: one and only commit is "Initial Checkin to prepare for automated 1.23 RC exports. This revision is just
• a copy of trunk - it is *NOT* the export.
• "
• [14:58] Soft Linden: CG cranks through JIRAs like a mad man. So he'd definitely see it there.
• [14:58] Rob Linden: Soft: good point...I think it's assigned to him already, but I'll check the status
• [14:58] Latif Khalifa: the svn sync needs to be solved in some systemic way
• [14:58] Rob Linden: CG has some great scripts that are that systemic way, however.....
• [14:58] Latif Khalifa: i think its silly to create a jira every time svn gets behind
• [14:58] Rob Linden: ....if a build fails, the push fails, I think
• [14:59] Rob Linden: ...and that may be what's going wrong
• [14:59] Robin Cornelius: oh, this also means the testing of the vertex object crash was a waste of time, as i don't actualy have 1.23 to compare with
• [14:59] Q Linden: latif...if by "systemic" you mean "shooting it in the head and hiding the body", yes :)
• [14:59] Youri Ashton: then that might be the problem with 64bit with vista and SL then
• [15:00] Youri Ashton: my machine usually doesnt get on SL, crashes etc for no reason. might bea simular problem
• [15:00] Rob Linden: Robin: ack.....that may be why I thought you had the source
• [15:00] Rob Linden: hrmph...ok. VERY sorry about that
• [15:01] Rob Linden: ugh....well
• [15:01] Youri Ashton: its okay, aslong as i dont loose my work inworld :p
• [15:01] Latif Khalifa: Q, nightlies are out for a few days, so I guess there were some successful 1.23 builds that would warrant a svn export :)
• [15:01] Youri Ashton: i just hope the subject will get more attention now since more people seem to have the same issues
• [15:01] Rob Linden: we know that CG and Merov are working together to get http-texture and 1.23 merged. that will effectively solve the problem. then the technicality of actually publishing just the 1.23 source remains
• [15:02] Q Linden: we should get another PN out tomorrow, btw
• [15:02] Youri Ashton: hi kormak
• [15:02] Kormak Blackheart: hello
• [15:02] Latif Khalifa: Q, so some builds are succeeding after all :)
• [15:02] Rob Linden: ok...well, we're out of time here today
• [15:02] Youri Ashton: wb merov!
• [15:02] Youri Ashton: and he is off again
• [15:02] Youri Ashton: lol
• [15:03] Rob Linden: Latif: the internal build process and thepublic build process work slightly differently
• [15:03] Rob Linden: it's possible for an internal build to succeed, but the public build to fail, because we use a slightly different method to build
• [15:03] Youri Ashton: brb
• [15:03] Latif Khalifa: still most interest is concentrated around next official viewer, having an up to date 1.23 branch would be nice :)
• [15:04] Rob Linden: by the way, I'm off on vacation next week
• [15:04] Rob Linden: Q will be filling in on this meeting, and the other Lindens on IRC can generally field the stuff that I otherwise might
• [15:05] Arawn Spitteler: hopes the other lindens can all manage a complete changeover of the system, to utterly confuse Rob
• [15:05] Aimee Trescothick: I thought you guys were only allowed virtual holidays?
• [15:05] Youri Ashton: bk
• [15:05] Rob Linden: anyway, we shoudl wrap this up...thanks everyone!
• [15:05] Rob Linden: Aimee: I'm breaking the rules
• [15:05] Harleen Gretzky: take care all
• [15:05] Aimee Trescothick: :D
• [15:05] Youri Ashton: thank you!
• [15:05] Robin Cornelius: Bye all
• [15:05] Robin Cornelius: thanks
• [15:05] Dirk Talamasca: No, they get a real holiday with virtual pay
• [15:05] Youri Ashton: bye bye :)
• [15:05] Aimee Trescothick: bye :)