Open Source Meeting/2008-05-22

< Open Source Meeting

Agenda

• Update from the Lindens (standing item) - Rob Linden
• Follow on from last week, why textures take so long to download, whats going on server side, Rob did your pinging find anything out? Michelle2 Zenovka 08:29, 22 May 2008 (PDT)
• General state and progress on OpenAL inclusion. Michelle2 Zenovka 08:24, 22 May 2008 (PDT)
• Some other issues with patches attached that may need a ping Michelle2 Zenovka 08:24, 22 May 2008 (PDT)
  • VWR-5697 Patch was attached some time after imported
  • VWR-2683 Defensive code, i'm still applying (Nicholaz closed himself) is this still valid?
  • VWR-2003 Defensive code, i'm still applying (Nicholaz closed himself) is this still valid?
• AWG_Test_Harness --basis for thin client? Saijanai 21:12, 24 January 1931 (PST)

Transcript

• [14:05] Rob Linden: hi folks, sorry for being late
• [14:05] Harleen Gretzky: Hi Rob
• [14:05] Carjay McGinnis: Hello Rob
• [14:07] Rob Linden: as usual, we plan to publish the transcript of this meeting, so please be aware of that
• [14:07] Rob Linden: http://wiki.secondlife.com/wiki/Open_Source_Meeting/Agenda
• [14:07] Jessica Kabumpo: signs the release forms
• [14:07] Rob Linden: first item: updates from Lindens
• [14:08] Rob Linden: I've done a little bit of work this week trying to improve the patch import process in general
• [14:08] Rob Linden: still in the early stages of this, but I'm pretty optimistic we'll be able to figure out how to make this better
• [14:09] Rob Linden: I'm also working to do a little more planning around how we use PJIRA, as there's more interest at Linden Lab in using it for collaboration with you all
• [14:10] Rob Linden: that's it for my update. others want to give updates?
• [14:11] Jessica Kabumpo: has a few issues
• [14:11] Q Linden: Notifications is proceeding, we're getting to the point of wanting to integrate with our latest releases
• [14:11] Asriazh Frye: hello everyone ^^
• [14:11] Q Linden: I've also been having conversations around the pyogp project, a python-based collection of utilities for testing protocols
• [14:12] Saijanai Kuhn: stuck a reference to that in the agenda
• [14:12] Saijanai Kuhn: test harness
• [14:12] Rob Linden: Jessica: make sure you put what you want to talk about on the agenda: http://wiki.secondlife.com/wiki/Open_Source_Meeting/Agenda
• [14:12] Liana Linden: My update: I've been doing a lot of internal training and process documentation around improving our open source communications and projects.
• [14:13] Jessica Kabumpo: growls
• [14:13] Liana Linden: Oh, and CMake is merging with main now. W00t!
• [14:13] Q Linden: w00ty toot t00t
• [14:13] Carjay McGinnis: woohoo
• [14:13] Qarl Linden: woot!
• [14:13] Michelle2 Zenovka: whooo
• [14:15] Rob Linden: k....any other updates?
• [14:16] Michelle2 Zenovka: i'm moving my viewer svn to git on Debians Alioth servers
• [14:17] Rob Linden: interesting.....dvcs is all the rage with the kids these days ;-)
• [14:17] Michelle2 Zenovka: i think if you could get in to it it would suit your development process
• [14:17] Q Linden: Well, we're leaning toward a distributed VCS, but it'll be Mercurial
• [14:17] Rob Linden: Bos Linden literally wrote the book on Mercurial
• [14:18] Carjay McGinnis: git is well suited for lots of parallel branches
• [14:18] Carjay McGinnis: hm, not done anything with Mercurial yet
• [14:18] Rob Linden: http://hgbook.red-bean.com/hgbook.html
• [14:18] Q Linden: The author works at LL. :-)
• [14:19] Rob Linden: oops, not bos linden....sardonyx
• [14:19] Rob Linden: anyway...
• [14:20] Rob Linden: any other updates, or shall we talk about the next item (which sadly I failed to follow up on)
• [14:20] Carjay McGinnis: book seems a good bed time read, thanks
• [14:20] Asriazh Frye: btw.. not using gcc 3.4 results in the moon not rendering.. in case noone noticed yet
• [14:21] Kooky Jetaime: Roflol
• [14:21] Saijanai Kuhn: note to enus: always test multiple time settings for QA
• [14:21] Carjay McGinnis: hm, I have it set to midday most of the time
• [14:22] Rob Linden: "Follow on from last week, why textures take so long to download, whats going on server side, Rob did your pinging find anything out? Michelle2 Zenovka 08:29, 22 May 2008 (PDT) "
• [14:22] Rob Linden: kicks himself
• [14:22] Michelle2 Zenovka: Nak gcc 4.2 and i have a moon
• [14:23] Q Linden: thinks it's not hard to find moons in SL
• [14:23] Carjay McGinnis: well, Qarl said he might be able to take a look if he found the time
• [14:23] Qarl Linden: he did?
• [14:23] Kooky Jetaime: Does the texture console have a wiki entry?
• [14:23] Asriazh Frye: i tried with 4.2.. no moon. maybe its the source i used but it has been like that for me for quite a while *shrugs*
• [14:23] Carjay McGinnis: *looks at Qarl*
• [14:23] Qarl Linden: he makes things up.
• [14:23] Rob Linden: so.....Qarl, this may be a good time to skim last week's transcript
• [14:23] Kooky Jetaime: checks his inventory.. Moon, Moon... Wheres my moon.
• [14:23] Rob Linden: http://wiki.secondlife.com/wiki/Open_Source_Meeting/2008-05-15
• [14:23] Qarl Linden: actually - i'm behind on my other work - so the texture stuff is on hold for the next couple weeks.
• [14:24] Qarl Linden: Michelle - does you patch help texture download speed?
• [14:24] Rob Linden: starting at the discussion of VWR-1815 VWR-2404
• [14:24] Qarl Linden: we *think* 2404 is now fixed.
• [14:24] Michelle2 Zenovka: Yes it does
• [14:24] Carjay McGinnis: BTW, I see the moon
• [14:25] Michelle2 Zenovka: the fix for 2404 still has a couple of textures that seem to not get to full discard
• [14:25] Michelle2 Zenovka: but its a great help
• [14:25] Qarl Linden: really?
• [14:25] Carjay McGinnis: gcc 4.1.3
• [14:25] Qarl Linden: can you send them to me?
• [14:25] Asriazh Frye: might be an ubuntu thing.. oh well...
• [14:25] Michelle2 Zenovka: I don't know if its unrelated but some linden trees were not showing there bark
• [14:25] Carjay McGinnis: it's Kubuntu Gutsy
• [14:25] Qarl Linden: hm.
• [14:26] Asriazh Frye: there go all my theories ^^
• [14:26] Michelle2 Zenovka: It might be unrelated
• [14:26] Michelle2 Zenovka: The download speed though it looks clear to me what is happening server side
• [14:26] Qarl Linden: the server stops sending data once it gets past the 8:1 level?
• [14:27] Qarl Linden: that should be pretty easy to fix on the server, i think.
• [14:27] Carjay McGinnis: that was the big question, does the server use the same rule as the client
• [14:27] Michelle2 Zenovka: yea, looks like for each requested discard the calc is used, should be a parallel fix to your viewer side fix
• [14:27] Qarl Linden: if (discard == 0), send all data.
• [14:27] Qarl Linden: yeah.
• [14:27] Carjay McGinnis: that's what i would expect, too
• [14:27] Qarl Linden: kk. as soon as a i get a moment, i'll check it out.
• [14:27] Michelle2 Zenovka: cool
• [14:27] Carjay McGinnis: thanks, Qarl, that would be great
• [14:28] Qarl Linden: if i fix it on the server - does that mean the patch you sent me isn't needed?
• [14:28] Michelle2 Zenovka: yes
• [14:28] Qarl Linden: cool.
• [14:28] Michelle2 Zenovka: its a pure work around
• [14:28] Kooky Jetaime: What does "NET" Mean on the State, under the texture console?
• [14:29] Qarl Linden: HEH.
• [14:29] Qarl Linden: i honestly don't use the texture console... so.
• [14:29] Asriazh Frye: theres this QAMode at the debug menu lately. which test features does it enable?
• [14:29] Michelle2 Zenovka: its the inital network request before it kicks to SIM
• [14:30] Michelle2 Zenovka: read the state machine in lltextureconsole.cpp
• [14:30] Kooky Jetaime: Thanks Michelle, I'll update the wiki
• [14:30] Rob Linden: ok, are we ready to move on? sounds like Qarl has got this one
• [14:30] Michelle2 Zenovka: Check the source to be sure *exactly what it is doing
• [14:31] Kooky Jetaime: *Looks at Michelle, ok, I won't update the wiki
• [14:32] Rob Linden: having the info on the wiki can't hurt, with of course the disclaimer that if the wiki and the source disagree, well, we all know which one is "right"
• [14:32] Rob Linden: next up: General state and progress on OpenAL inclusion. Michelle2 Zenovka 08:24, 22 May 2008 (PDT)
• [14:32] Kooky Jetaime: yea, cept those reading the wiki would probably never realize that it conflicts
• [14:32] Michelle2 Zenovka: Ok it initalises the fetcher ready to recieve packets in the SIM state
• [14:34] Rob Linden: re: OpenAL. we're still working on it. I can't go into too much detail other than to say we've potentially got some third party help that we may be getting
• [14:34] Rob Linden: ...and to say that we're still sorting out how Ogg/MP3 streaming will work
• [14:34] Rob Linden: ...on Windows and Mac
• [14:34] Michelle2 Zenovka: ok, but i would have not though that we are that far off the basic implementation (excluding streaming)
• [14:34] Liana Linden: /brb
• [14:34] Michelle2 Zenovka: um have you seen openal-soft ?
• [14:35] Tegg Bode: You can sit up here Jessica :)
• [14:35] Michelle2 Zenovka: its got streaming capabilites but needs a decoder plugged in
• [14:35] Jessica Kabumpo: was showing the bug she discovered
• [14:36] Tegg Bode: Oh oops :)
• [14:36] Rob Linden: the biggest issue with us making it the default is making sure that we have reliable support for MP3 and Ogg on all three platforms. the great thing about Ogg Vorbis is that we can ship that with the viewer, so that's not a big concern, but with MP3, we still have to figure out what our strategy is
• [14:37] Rob Linden: anyone here have experience with GStreamer on Windows/Mac?
• [14:37] Michelle2 Zenovka: no, may give it a go. I expect it to *just* work as its Gdk based
• [14:37] Rob Linden: heh
• [14:38] Rob Linden: multimedia doesn't just work cross platform :)
• [14:38] Michelle2 Zenovka: binaries avaiable
• [14:38] Rob Linden: if by emphasizing "just", you mean "barely", I'd go with that assumption :)
• [14:38] Michelle2 Zenovka: hehe
• [14:39] Rob Linden: I'm encouraged by this, though: http://blog.songbirdnest.com/2008/01/30/gstreamer-for-all-all-for-gstreamer/
• [14:39] Liana Linden: /back.
• [14:39] Michelle2 Zenovka: i presume the core is good but codecs are an issue
• [14:39] Michelle2 Zenovka: also patent lumbered
• [14:39] Rob Linden: (and this): http://blog.songbirdnest.com/2008/03/26/gstreamer-progress/
• [14:39] Michelle2 Zenovka: hey, i look at work and see what happens
• [14:40] Michelle2 Zenovka: don't have windows systems at home to try
• [14:41] Rob Linden: I wouldn't assume anything about the core, though I'll grant you that things get much easier when dealing with only audio rather than a/v
• [14:41] Jessica Kabumpo: has her iMac named Cougar that she runs SL on
• [14:42] Rob Linden: anyway, progress is happening on OpenAL, but we don't yet have the bandwidth to cram it in the queue of stuff that NEEDS to go out
• [14:43] Rob Linden: ....and it's a big enough change that we can't just bundle it in as a minor feature
• [14:43] Michelle2 Zenovka: it could in theory go in and be swithed out by compile flags
• [14:43] Michelle2 Zenovka: if you had the man power
• [14:43] Michelle2 Zenovka: but hey its progress ;-)
• [14:43] Saijanai Kuhn: open source licensing issues there?
• [14:44] Rob Linden: Saijanai: it's the whole package (licensing, testing, project managment, etc)
• [14:44] Q Linden: michelle2, IMO "in but switched off" is worse than "not in" for maintenance reasons
• [14:45] Carjay McGinnis: think Soft once explained that compile flags was a no go because it would needed to be tested separately in this case for all platforms etc. all the time
• [14:45] Saijanai Kuhn: just thinking that having compile options to support two different libraries in teh same open source project might be a bit iffy
• [14:45] Q Linden: exactly
• [14:45] Rob Linden: let's move on - I promise, we're making progress on this one
• [14:45] Rob Linden: mini patch triage now:
• [14:45] Saijanai Kuhn: if you are using this platform, use this license exception but if you are using this platform, use this license, EXCEPT if...
• [14:45] Rob Linden: VWR-5697[c
• [14:46] Rob Linden: https://jira.secondlife.com/browse/VWR-5697
• [14:46] Michelle2 Zenovka: yea i think i got to the bottom of this
• [14:47] Michelle2 Zenovka: there is a default initalisation for a patch that is the same as the binary location, and it tries to be written to, which UAC will hate
• [14:47] Michelle2 Zenovka: *path
• [14:47] Rob Linden: oh...so the original repro case was too vague, but you narrowed it down?
• [14:48] Michelle2 Zenovka: I think so, i found a bug for sure, But don;t have vista to be sure its 100% the bug reported
• [14:49] Michelle2 Zenovka: there is certainly no way volume_Settings and url_history should be written to at the binary path, this only gets updated once a sucessful login as happened
• [14:50] Rob Linden: k....I'll put it in the comments
• [14:50] Rob Linden: (on the internal counterpart)
• [14:51] Rob Linden: next up: https://jira.secondlife.com/browse/VWR-2683
• [14:51] Michelle2 Zenovka: These are just two old defensive patches, i still apply them but unneeded or worth the effort?
• [14:52] Michelle2 Zenovka: They were a problem at some point
• [14:53] Michelle2 Zenovka: just pinging the issues to either kill for good or what ever
• [14:53] Rob Linden: I don't have a strong opinion on them
• [14:53] Rob Linden: anyone else care to comment?
• [14:53] Michelle2 Zenovka: I presume there was another bug that was triggering the events these protect?
• [14:54] Michelle2 Zenovka: and thats probably long gone as it would have been an issue
• [14:54] Carjay McGinnis: hm, that was before my time, don't know anything about it
• [14:55] Rob Linden: I have no idea. they're still marked as open in our internal tracker, and QA did take a look at a repro for the first one
• [14:56] Michelle2 Zenovka: thats probably all we can do then
• [14:56] Rob Linden: yup...the second one is assigned to Soft. I'm going to put it back in the triage queue
• [14:57] Rob Linden: k...that was the agenda
• [14:57] Rob Linden: any other quick items before we go?
• [14:57] Liana Linden: Thanks for agenda-wrangling, Michelle2.
• [14:57] Saijanai Kuhn: except what I put on there
• [14:58] Saijanai Kuhn: just a heads up about the test harness page and discussion
• [14:58] Michelle2 Zenovka: Its vanished?
• [14:58] Michelle2 Zenovka: I saw it earlier
• [14:58] Saijanai Kuhn: still there I just refreshed
• [14:58] Rob Linden: looks
• [14:59] Michelle2 Zenovka: Heh not on mine
• [14:59] Saijanai Kuhn: we had a meeting at Which's office hours. Tao, Enus, Which, and I. Just chatted about where it was going and how it would work
• [14:59] Rob Linden: weird...I seem to be having a caching issue or something
• [14:59] Saijanai Kuhn: https://wiki.secondlife.com/wiki/AWG_Test_Harness
• [14:59] Q Linden: yeah, I couldnt' see it either
• [14:59] Carjay McGinnis: not getting it either
• [15:00] Saijanai Kuhn: very odd. refershed and still there
• [15:00] Saijanai Kuhn: though it did disappear at one point.
• [15:00] Saijanai Kuhn: regardless it was just a pointer back to teh test harness page
• [15:00] Rob Linden: ok....I think there's a caching issue on one of the wiki mirrors
• [15:00] Rob Linden: fyi, the workaround if you see that happen is to tack ?action=purge at the end of the url
• [15:00] Carjay McGinnis: ah, now I get it, too
• [15:01] Saijanai Kuhn: AWG_Test_Harness --basis for thin client? Saijanai 21:12, 24 January 1931 (PST)
• [15:01] Saijanai Kuhn: I just noticed the date and time too..
• [15:01] Jessica Kabumpo: Lawl ^-^
• [15:01] Carjay McGinnis: old one
• [15:01] Jessica Kabumpo: very old
• [15:02] Saijanai Kuhn: that was a ~~~~ signature
• [15:02] Rob Linden: ugh
• [15:02] Rob Linden: ok.....looks like there's some sort of time problem
• [15:02] Tegg Bode: hmm is that the B&W version of Beta viewer 0.0?
• [15:02] Carjay McGinnis: could explain the refresh problems, page set to being older
• [15:03] Saijanai Kuhn: anyway just a link to the test harness page. The framework we talked about should be useable for both testing new protocols and for using old ones as well (for thin client)
• [15:03] Jessica Kabumpo: I'm going elsewhere, but ... my issue .... think of BSD?
• [15:03] Saijanai Kuhn: eh, the proposal for the test client was Apache license, I think
• [15:03] Rob Linden: ok....well, we're running over, and some of you may want to go to Ben's office hour
• [15:03] Jessica Kabumpo: heads to Ben's
• [15:04] Michelle2 Zenovka: rats that was an image decode crash
• [15:04] Liana Linden: Jah. Gotta head. Thanks.
• [15:04] Rob Linden: thanks everyone!
• [15:04] Carjay McGinnis: a crash in OpenJPEG? Ugh
• [15:04] Q Linden: thanks
• [15:04] Saijanai Kuhn: laters all
• [15:04] Carjay McGinnis: thank you, Rob
• [15:04] Tegg Bode: Yep gtg, You can beam me up now Phillip