Open Source Meeting/2008-11-06

Agenda

• Vivox and slvoice, for example VWR-4589 is fixed in the version of SLVoice shipped with the SLIM first look, no one knew this, vivox say its LL who decided the release schedule. How can we push these fixes faster or propagate the info. Some of us are spending time trying to create work arounds and solutions for issues that no-one knows is fixed. (Robin Cornelius)

• Webkit news? (Robin Cornelius)

Transcript

• [14:00] Carjay McGinnis: Jodiah: https://lists.secondlife.com/pipermail/sldev/2008-October/012065.html
• [14:00] Jodiah Jensen: wow Carjay, thanks!
• [14:00] Robin Cornelius: rats too lagy brb in a differnt client
• [14:01] Geneko Nemeth: Translation is almost complete...
• [14:01] Geneko Nemeth: Wheep!
• [14:01] Rob Linden: hi folks
• [14:02] Jodiah Jensen: Ahoy Rob
• [14:02] Squirrel Wood: Hollas!
• [14:02] Rob Linden: transcript of this meeting will be posted later on wiki.secondlife.com, so please don't post anything you don't want to see there
• [14:02] Rob Linden: er...rather don't type anything here that you don't want to see there
• [14:03] Squirrel Wood: The cake is a pie. You can see that in its sauce code! :p
• [14:03] Carjay McGinnis: huhu Squirrel :)
• [14:03] Soft Linden: badum ba
• [14:03] Rob Linden: http://wiki.secondlife.com/wiki/Open_Source_Meeting/Agenda
• [14:04] Rob Linden: First item: Vivox and slvoice, for example VWR-4589 is fixed in the version of SLVoice shipped with the SLIM first look, no one knew this, vivox say its LL who decided the release schedule. How can we push these fixes faster or propagate the info. Some of us are spending time trying to create work arounds and solutions for issues that no-one knows is fixed. (Robin Cornelius)
• [14:04] Robin Cornelius: yea this came about as we spent a week messing about trying to override libc open() read() to fix a know vivox error
• [14:04] Squirrel Wood: Is there a vivox/voice meta issue where this kind of information could be posted ?
• [14:05] Robin Cornelius: then asked on vivox open and they said its already fixed and in the new firstlook
• [14:06] Rob Linden: I'm paused here a little bit because I'm tryign to think through how to improve the process
• [14:06] Soft Linden: There is an interenal list of bug fix asks. I'm wondering if that could get a column of pjiras to update
• [14:07] Soft Linden: How often are these things an issue? If this took a lot of manual involvement from a PM, would there be a good case for it still?
• [14:07] Soft Linden: Or would it be better to go case by case - if you're chasng down a voice issue, we could escalate that to the team when it comes up if you mention it on sldev?
• [14:08] Robin Cornelius: well there are a stream of issues with the vivox client, if there was a latest version somewhere seperate to the main vieweer distros that could be helpful
• [14:08] Soft Linden: Fair point. You shouldn't have to watch every last branch to discover the newest version.
• [14:08] Robin Cornelius: seperate to that i'm discussing with vivox how to do a "nvidia gfx driver" style non free distribution
• [14:08] Rob Linden: definitely bring these things up more on sldev. for that matter, I don't think it hurts to ask on vivox-open, even if that's not the official channel.
• [14:08] Jodiah Jensen: I would think if something gets fixed, it would be a "good thing"(tm) to announce it somewhere, definitely, but we're relying on a third-party developer to announce their fixes in this case, no? isn't vivox/slvoice one of the non-license-compatible libs?
• [14:09] Robin Cornelius: vivox are saying its LLs job to anounce and scheule releases
• [14:09] Geneko Nemeth: Technically, it's a separate application...
• [14:09] Jodiah Jensen: but how can LL announce it if Vivox dosn't tell LL? lol
• [14:09] Rob Linden: Robin: correct....I just reread Jim's response
• [14:10] Rob Linden: what I'm saying is that Vivox can say "this is fixed in our latest SDK....you'll have to ask Linden Lab when the fix is going to make it in".
• [14:10] Rob Linden: (which is in fact what Jim @ Vivox said)
• [14:10] Jodiah Jensen: ah I see
• [14:11] Squirrel Wood: So it would be possible to update a supposed umbrella/meta issue ?
• [14:11] Rob Linden: what I'm saying is that it doesn't hurt to just let the VIvox folks know that there issues out there and update them from time to time in a non-obnoxious way
• [14:12] Rob Linden: ultimately Vivox<->Linden Lab<->community is the "official" channel for the information, but these unofficial channels can help get the right people talking to each other
• [14:13] Rob Linden: that said, raising this stuff on sldev@ is welcome too
• [14:13] Rob Linden: to Squirrel's question: a meta issue couldn't hurt
• [14:14] Rob Linden: note that we do at least have the "Voice" component
• [14:14] Carjay McGinnis: well, that's what we did and they said we should put things on JIRA
• [14:14] Rob Linden: http://jira.secondlife.com/secure/IssueNavigator.jspa?reset=true&pid=10003&resolution=-1&component=10050&sorter/field=summary&sorter/order=ASC&sorter/field=resolution&sorter/order=ASC&sorter/field=status&sorter/order=ASC&sorter/field=priority&sorter/order=DESC
• [14:15] Carjay McGinnis: I did see a comment from vivox employees somewhere but not for that specific bug
• [14:15] Rob Linden: Carjay: right....you asked about what you should do generally, and they told you. file it in JIRA
• [14:16] Rob Linden: what I'm saying is that if there are specific things you want to discuss status on that are already filed in JIRA, it probably doesn't hurt to bring them up on vivox-open
• [14:16] Rob Linden: not that that is an official channel
• [14:17] Rob Linden: anyway, we'll also take a look at how to make the pipeline a little more efficient here at Linden Lab
• [14:18] Soft Linden: Still - an sldev recap would be the most helpful starting point. That would be useful to point the voice team at when starting the discussion.
• [14:19] Rob Linden: yup....a handy tip for working with us in general. the closer you can get things to a format that we can forward with a simple "any ideas?" annotation, the more likely you are to get a response :)
• [14:20] Jodiah Jensen: where does vivox announce updates to their SDK? is there a list?
• [14:20] Carjay McGinnis: yes, well, I thought they set up their list for that but it is not really used
• [14:20] Carjay McGinnis: in fact the only messages on it are from me, Robin and poppy I believe
• [14:21] Rob Linden: also, the more specific you are (e.g. what's the status on VWR-54321, the problem with x, y, and z?), the more likely we are to get an answer
• [14:21] Carjay McGinnis: and some responsed from vivox employees of course
• [14:21] Carjay McGinnis: responses
• [14:22] Rob Linden: so, asking about what the vivox-open list is for is a fine question for the vivox-open list :)
• [14:22] Rob Linden: anything else on this topic, or can we move on?
• [14:23] Rob Linden: moving on then: * Webkit news? (Robin Cornelius)
• [14:24] Rob Linden: so, we have a contractor hired who has been working on a couple of different approaches
• [14:24] Robin Cornelius: any successes?
• [14:24] Rob Linden: I'll need to check in to see what more I can talk about
• [14:26] Rob Linden: point of process for both of these items.....it's always best to ask in sldev@ prior to putting something on the agenda of this meeting. not that I don't appreciate showing up to see an agenda, but the odds are greater that I'll be able to actually say somethign useful in real time if I'm actually expecting the question
• [14:26] Robin Cornelius: yea was planning to but ran short on time
• [14:26] Carjay McGinnis: you don't like surprises, eh? :)
• [14:27] Carjay McGinnis: well, just kidding, understand what you are saying
• [14:27] Soft Linden: Not that the technical bits are going to stay secret or anything - but the business bits on this shape what can/should be said until they're ironed out and we're -told- what's safe. :)
• [14:27] Rob Linden: Robin: not a big deal
• [14:27] Robin Cornelius: well as for webkit, i've been experimenting and have a very very basic system working with a clutter backend
• [14:27] Rob Linden: having an agenda is better than not having an agenda
• [14:27] Robin Cornelius: that can render to a memory buffer
• [14:27] Rob Linden: sweet!
• [14:27] Robin Cornelius: using only libs avaiable on debian
• [14:28] Robin Cornelius: but integrating to a full mozlib library is more work, but in theory its only doing stuff rather than inventing things
• [14:29] Rob Linden: are you using ubrowser?
• [14:29] Robin Cornelius: no
• [14:29] Robin Cornelius: i could try to connect it to the ubrowser app
• [14:29] Rob Linden: the API should be the same
• [14:29] Robin Cornelius: i#m starting with a null molib
• [14:30] Robin Cornelius: i've got a blank mozlib with all functions stubbed out
• [14:30] Rob Linden: ....and that's how a lot of the development is done here
• [14:31] Rob Linden: Some other tech that we're looking at: QtWebkit, Skia and Awesomium
• [14:31] Rob Linden: there's a lot of different ways to skin this cat
• [14:31] Geneko Nemeth: Hey!
• [14:31] Robin Cornelius: i've been around this loop
• [14:31] Robin Cornelius: QT has offscreen rendering capability
• [14:32] Robin Cornelius: cairo currently does not (aka gtk ports of webkit)
• [14:32] Robin Cornelius: what about clutter?
• [14:32] Robin Cornelius: Skia is the google webkit engine API
• [14:32] Robin Cornelius: and Awesomium seems to be someone working with Skia
• [14:32] Rob Linden: it's been a while since we looked at Clutter
• [14:33] Rob Linden: yeah, Awesomium is a really thin layer on top of Skia
• [14:33] Robin Cornelius: now i guess the basic issues are the old dynamic updating
• [14:34] Robin Cornelius: some of the webkit ports are fairly basic and you need observers
• [14:34] Robin Cornelius: which is what callum hacked into xulrunner
• [14:35] Robin Cornelius: Can you say how successful any routes have been, have you seen tangable results?
• [14:36] Rob Linden: QtWebkit is going to be the most fruitful direction for short term win
• [14:36] Rob Linden: ....esp on Linux
• [14:36] Geneko Nemeth: But wouldn't that bring dependency on Qt?
• [14:37] Robin Cornelius: ack
• [14:37] Geneko Nemeth: With the custom preprocessor and stuff.
• [14:37] Rob Linden: Geneko: yes it would
• [14:37] Rob Linden: Robin: which aspect were you ack'ing about?
• [14:37] Robin Cornelius: dep on qt
• [14:37] Soft Linden: I don't think we'd let the custom preprocessor stuff escape the library build.
• [14:38] Geneko Nemeth: Especially that, we've already linked GTK+. XD
• [14:38] Rob Linden: Robin: yes, but what's the reason for the aversion to Qt?
• [14:38] Robin Cornelius: no particular reason, it just seems crazy linking Gtk and Qt,
• [14:39] Rob Linden: ah, yeah, that aspect would be unfortunate
• [14:39] Robin Cornelius: i'll probably carry on playing as i get time with my clutter attempt
• [14:39] Soft Linden: Been looking in the background here - didn't realize webkit-clutter was a mature as it seems
• [14:39] Rob Linden: doesn't recall what all Gtk+ is used for
• [14:40] Rob Linden: (in the viewer, that is)
• [14:40] Geneko Nemeth: To display file upload windows...
• [14:40] Robin Cornelius: webkit clutter is becomming an offical port
• [14:40] Soft Linden: Out of curiosity, why do you think Google wouldn't have gone the webkit-clutter route?
• [14:40] Rob Linden: (ooo....good question)
• [14:40] Geneko Nemeth: I'm not sure if there are any other uses since I didn't read the source ^_^||
• [14:40] Robin Cornelius: they've created there own port so they can get to the lower level aspects of webkit
• [14:41] Robin Cornelius: the ports tend to abstract too much away for a full browser like google were atempting
• [14:41] Soft Linden: Does webkit-clutter deal with plugins?
• [14:42] Robin Cornelius: i would guess not
• [14:42] Robin Cornelius: but then it might if its a webkit native thing
• [14:44] Soft Linden: Looks like someone's doing a NS plugin wrapper as a clutter plugin though.
• [14:44] Rob Linden: Robin: what sort of things get abstracted away?
• [14:44] Soft Linden: So embedded plugins might not, but they might still be accessible as full pages with a bit of hackery
• [14:44] Robin Cornelius: well the exposed API seems very simple, but then i've not really dug deep
• [14:44] Rob Linden: wonders what we would lose by using one of these higher level ports
• [14:45] Robin Cornelius: well thats a question a dev with time could find out farily shortly
• [14:45] Soft Linden: Might ask the contractors if they've been down this route, have any ideas?
• [14:45] Robin Cornelius: yea
• [14:46] Robin Cornelius: heh looks like i've given you something to think about
• [14:46] Rob Linden: mentioned that a webkit conversation was going on on our internal irc
• [14:47] Soft Linden: If clutter's really becoming an official webkit backend, that's huge though.
• [14:47] Robin Cornelius: yes
• [14:47] Rob Linden: what does "official" mean? downloadable from webkit.org?
• [14:47] Robin Cornelius: yes in the webkit git tree
• [14:48] Robin Cornelius: currently clutter is synced with webit git every few days i believe
• [14:48] Rob Linden: so does that mean the clutter devs commit common code directly to the master branch?
• [14:49] Robin Cornelius: i would think not directly, esp with git
• [14:49] Robin Cornelius: they might have a push branch
• [14:50] Rob Linden: I'm trying to figure out just how big a deal it is that clutter is "official", because being hosted by webkit.org isn't *always* going to be a big deal
• [14:50] Rob Linden: (e.g. being hosted on code.google.com doesn't imply endorsement by Google, for example)
• [14:51] Rob Linden: I'm sure being hosted on webkit.org is more significant than that, but it's a question of how much more significant
• [14:51] Robin Cornelius: it implys its not going to vanish overnight
• [14:51] Rob Linden: it'll just languish instead ;-)
• [14:52] Rob Linden: kidding of course. being backed by Intel now and being hosted on webkit.org is a good combo
• [14:53] Rob Linden: anyway, thanks for the update Robin, and I'll try to have a better answer on the webkit stuff on our side by the next meeting
• [14:53] Rob Linden: feel free to pester me on sldev about it :)
• [14:53] Robin Cornelius: sure
• [14:54] Rob Linden: well, there was nothing else on the agenda. anything else to talk about?
• [14:54] Robin Cornelius: almost time anyway
• [14:55] Carjay McGinnis: yeah, good timing
• [14:55] Soft Linden: thumbsup
• [14:55] Soft Linden: Thanks much for coming, all!
• [14:55] Rob Linden: thanks everyone!
• [14:55] Carjay McGinnis: thank you, too
• [14:55] Jodiah Jensen: thanks for your time :)
• [14:55] Geneko Nemeth: Nya ha ha.
• [14:55] Robin Cornelius: thanks Bye!
• [14:55] Geneko Nemeth: Seeya then.