Web   ·   Wiki   ·   Activities   ·   Blog   ·   Lists   ·   Chat   ·   Meeting   ·   Bugs   ·   Git   ·   Translate   ·   Archive   ·   People   ·   Donate

#sugar-meeting meeting, 2011-01-08 14:01:34

Minutes | Index | Today     Channels | Search | Join

All times shown according to UTC.

Time Nick Message
14:01 meeting Meeting started Sat Jan  8 14:01:34 2011 UTC. The chair is m_anish. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01 Useful Commands: #action #agreed #help #info #idea #link #topic #endmeeting
14:01 silbe Happy new year everyone!
14:02 m_anish silbe, happy new yr!
14:03 should we start with discussing what has everyone been up to and then move to py related topic?
14:03 dfarning m_anish, will be leading the meetings for the next couple of week - as we will be driven by the need of PyEduca for dextrose:)
14:03 silbe m_anish: sounds good to me. get back in sync and then start the open-ended topic.
14:04 m_anish silbe, okay, lets start with you :)
14:05 silbe As you've probably noticed, I announced the new review process Simon and I agreed on. Since there was no opposition, I consider it in place.
14:06 dfarning silbe, yes that was a huge step in the right direction.
14:07 silbe So we should be good now for "external" patches (as both Simon and I can ACK them); let's hope it works well for patches from Simon and me, too (since we need to ACK each others patches). ;)
14:09 m_anish silbe, sounds great! what are your plans for the next 1-2 weeks!
14:09 silbe Over the "holidays" I've worked on the Dextrose-3 patch set. Finally got wifi disconnect working yesterday. Quozl's patch wasn't enough, a recent question resp. the answer to it by somebody else on the NetworkManager list explained what we were doing wrong so I could now fix it.
14:09 bernie silbe: i'm very happy. this fixes a very longstanding problem
14:10 silbe There are a few outstanding issues, but most patches that I intend to land in mainline soon are in good shape now.
14:11 This doesn't include the accessibility series yet. I haven't made my mind up on it as I don't know enough about what exactly it's expected to solve and why we need to patch Sugar for that.
14:12 backup support is undecided as well
14:12 bernie silbe: tch is now in the right place to ask!
14:12 silbe: note that users in paraguay liked the accessibility patches very much
14:13 silbe bernie, tch: Great! I'll get back to you soon.
14:13 bernie silbe: especially the virtual keyboard and the fonts (for teachers who can't see well)
14:13 m_anish just curious, is there a need for separate branches for the patches we intend to upstream and those we don't?
14:13 silbe the plan for the near future is to continue working on the Dextrose-3 patch set#
14:14 m_anish: we already have that branch: mainline/master. <egdr>
14:16 bernie: there's a bit of confusion around virtual keyboards (what solutions are out there, who uses them, etc.). We should talk about that later, too.
14:17 Collaboration is pretty broken in 0.90/0.91, BTW. I expect that it will take a specialist a few weeks full-time to clean it up. :-/
14:17 bernie silbe: yes, ideally we'd want something properly integrated with the gnome accessibility stack... but I don't think we have the resources to do the proper thing
14:18 silbe bernie: actually I'd hope that we could just use something the Gnome people already wrote. But let's talk about that after the meeting.
14:18 bernie silbe: without sayamindu, the current virtual keyboard, as limited as it might be, is probably going to be the best thing we have for a long time.
14:18 silbe *hoped
14:18 bernie silbe: oh, there's something we could reuse? GURAIT-TO!
14:19 silbe: for collaboration, perhaps quozl would be interested?
14:19 silbe bernie: it's just a hope. I don't know what actually exists.
14:20 m_anish silbe, bernie +1 for after-meeting discussion :), tch can you let us in on your uy work :)
14:20 silbe bernie: has he worked with Telepathy before? I spent some time myself debugging, but without knowledge of how Telepathy works, it's quite hard to figure out.
14:23 let's talk about that later as well and continue with the meeting. I already took much more time than I wanted to. :)
14:23 bernie silbe: I'm probably wrong... he's been hacking a lot on *networking*, not collaboration
14:23 tch m_anish: of course, I will spend the next week at the plain ceibal offices. My main goal is to help them with their dextrose bug reports (since they have been doing real testing lately)..
14:23 plan*
14:24 m_anish: there are also more things i could do there, like convince them to stick with dextrose2 as close as possible ;)
14:26 m_anish tch, do you think there are patches in uy that py dxo-2 builds could use?
14:27 alsroot m_anish: any progress w/ gathering items for TODO-temp?
14:27 tch m_anish: that is something i will find out there :)
14:27 bernie tch: fantastic
14:28 alsroot m_anish: tch was talking also about auto bugs reporting
14:28 m_anish alsroot, alas, not much, i was supposed to get the list yesterday from rralcala (roberto), but didn't :/
14:28 alsroot, mukesh_dce and tch are exploring that :)
14:28 tch yeah that is an important topic IMHO..
14:29 I am trying to figure out if we can use ABRT
14:29 silbe tch: BTW, has there been any progress on your notifications stuff?
14:29 bernie alsroot: I just hope we don't end up like this: http://lwn.net/Articles/422062/
14:29 tch silbe: yeah, i did not updated the feature page but i think i got a design (implemented) that won't hurt anyone haha
14:30 alsroot bernie: in what case?
14:30 m_anish silbe, fyi, we have the notification stuff integrated in dx2-py builds
14:31 bernie alsroot: for receiving bug reports and ignoring them
14:31 tch bernie: i think receiving them is a great step forward
14:32 alsroot bernie: I was thinking about simple HTTP POST reports that need to be sorted by dextrose team
14:32 m_anish bernie, alsroot tch if and when we have abrt up, we'll need someone to triage. without it, it wouldn't make much sense
14:32 alsroot ..like it weas done in Log
14:32 bernie tch: as long as we don't give the users the false impression that we're going to do something about it
14:32 tch: because, currently, we've not been working on fixing any of these bugs: http://bugs.sugarlabs.org/wiki/Dextrose
14:32 silbe m_anish: hmm. I'm a bit worried that by exposing users to the current, not-yet-agreed-on design, we're forcing upstream to keep using it because users will complain about UI changes later. Similar to how OLPC is forcing us to keep supporting their new metadata storage format for external media...
14:33 bernie alsroot: I think mtd was working on something like that
14:33 or wadeb
14:34 tch bernie: well, i think we are building our ground to start tackling some of them, :)
14:34 bernie alsroot: http://logcollect.sugarlabs.org/
14:34 (broken)
14:34 alsroot m_anish: tch: anyway, we need to have some kind list to duscuss priorities (and level of impls), m_anish: could you ping us (/me and tch) when you get new items
14:34 tch bernie: but i understand your point and its valid.
14:34 bernie tch: cool
14:34 m_anish alsroot, +1 will ping
14:35 silbe something abrt-like would probably be OK, as it only collects segfaults (which are almost always rather severe bugs).
14:35 alsroot bernie: ^, particualar bugs reporting might be pretty different, I'd prefer to implement as simple as possible and accoring to the TODO items for 1 Feb dead line
14:35 bernie silbe: do you have more info on this metadata thing?
14:35 silbe: I had overlooked it
14:35 m_anish silbe, i think we're just in the planning stages re: abrt and ther would be a mail to sugar-devel before we start to write code
14:35 tch, what do you think?
14:35 silbe as for easier reporting of other bugs, I too am worried about us drowning in reports. We already can't keep up with the existing reports.
14:36 m_anish: the nice thing about the abrt stuff is that it's invisible to the user, so upstream doesn't need to care about it.
14:36 bernie silbe: yup. we need someone to act as PM and start dispatching the bugs to developers for fixing them
14:37 alsroot m_anish: for me abrt is looking overkill, we might need only python logs, but once more, it depends on TODO for 1 Feb
14:37 silbe bernie: just take a look at the (somewhat) recent changes in sugar/mainline/sucrose-0.84
14:38 bernie silbe: in fedora, abrt has a UI and requires several (easy) steps to file a bug. like, for example, describing the context and approving the traceback for privacy issues
14:38 silbe bernie: hmm.
14:38 bernie silbe: so 0.84, like dextrose, contains non-upstreamed patches now?
14:39 m_anish bernie, +1 for PM needed
14:39 bernie silbe: not long ago, dsd was convinced that it wasn't the case
14:41 silbe bernie: It seems so, yes. Though technically it's "upstream", but not in the latest development version, which makes it even worse.
14:41 E.g. Debian packages new 0.84 each time Simon pushes one out.
14:42 tch bernie: should we try to get some consensus about the automatic bug reporting?
14:42 bernie silbe: maybe we should tell erikos not to do it then
14:43 silbe: he can create a 0.84-olpc branch if he wants to
14:43 SMParrish_mobile <SMParrish_mobile!~Android@> has joined #sugar-meeting
14:44 bernie tch: I'm in favor, but not now. it's a lot of work to integrate it, and we already don't have enough resources to fix all the bugs already in trac
14:44 tch: also, see this: http://bugs.sugarlabs.org/ticket/1439
14:45 tch: there was already some work done, but not with abrt
14:45 m_anish bernie, +1, tch maybe keep it as low priority (time wise)
14:46 tch bernie: i guess that is what alsroot was thinking about
14:47 silbe tch, bernie: how about doing "debug" builds that automatically report everything to a build-time configured server (=> deployments can use their own server and forward to us). No UI and no privacy issues for regular users either.
14:47 m_anish alsroot, can you sync us up on ur work of past few weeks :), then I guess we can discuss py for a bit :)
14:48 bernie silbe: can we reuse wadeb's patches for logcollect?
14:48 silbe: it seems easier to integrate than abrt to me
14:49 alsroot is working on http://wiki.sugarlabs.org/go/P[…]form_Team/Roadmap plan, it doesn't relate directly to upcomming dextrose release
14:49 bernie silbe: for example, abrt supports only bugzilla, we'd have to write a plugin for trac
14:49 alsroot, silbe: are the two of you coordinating on these upstream efforts?
14:49 silbe bernie: I would have expected abrt to be able to just run in daemon mode with minimal configuration. But maybe I'm expecting too much of it...
14:49 alsroot bernie: for bugs reporting case?
14:50 tch silbe: i was expecting the same..
14:50 bernie silbe: yes, it does... but then it needs to bring up a UI to file the actual bugs.
14:50 silbe bernie: if you're talking about http://wiki.sugarlabs.org/go/P[…]form_Team/Roadmap , then no, I still don't get this "sweets" stuff.
14:50 bernie silbe: it needs also packagekit to download the debuginfos
14:50 SMParrish_mobile Silbe  abrt is a work in progress  i can talk to the devs at fudcon and see about trac support
14:50 silbe bernie: so it can't just dump a tarball on some server?
14:51 SMParrish_mobile: hey, nice to see you!
14:51 tch SMParrish_mobile:  :)
14:51 alsroot bernie: in my mind pushing it to upstream is too early, we need to test it in the field and upcomming dextrose release is a good chance
14:51 SMParrish_mobile thanks  glad to be back
14:51 m_anish silbe, it should be able to, otw, we don't use it :)
14:51 silbe SMParrish_mobile: the simpler the better I'd say. No Trac or other special tool, just HTTP upload or email.
14:51 bernie SMParrish_mobile: cool. our usecase is slightly different from the gnome desktop one. we want bug reports to be collected silently, and we might not need to download debuginfos, as we mostly have only python code.
14:52 m_anish SMParrish_mobile, hola!
14:52 bernie alsroot: would would have to be done?
14:52 m_anish alsroot, btw, we're not targeting abrt for dx2 feb-1 release
14:52 alsroot m_anish: for what date then, dfarning was talkink about 1 Feb
14:52 bernie alsroot: dextrose2 is due in ~2 months, I'm afraid we missed this train
14:52 silbe bernie: isn't abrt just for segfaults? Those happen at C level and need debug infos.
14:53 bernie silbe: it also does python tracebacks
14:53 silbe bernie: which would be another argument pro special "debug" builds
14:53 m_anish alsroot, abrt is a good to have feature, not a necessity for dx2-py initial release IMHO
14:53 bernie silbe: I actually wonder how it notices about those for processes running independently
14:53 m_anish Can I start to discuss a bit abt py-roadmap?
14:53 silbe bernie: interesting. All ones or just those that abort the process?
14:53 bernie m_anish: yup
14:54 alsroot m_anish: in my mind abrt is overkill, if we are getting core dumps it is something too bad, and most likely w/ sys libs not for our code
14:54 bernie silbe: I think the latter
14:54 silbe m_anish: sorry. We'll discuss abrt later.
14:54 dfarning SMParrish_mobile, will restart as PM.  he took some time off for family reasons.  We will work together next week to ensure a smoother hand off in case of unexpected absences.
14:54 SMParrish_mobile Bernie   abrt is good at catching crashes in c code   not sure hiw good the python support is atm
14:54 alsroot m_anish: just posting activity logs via HTTP should be pretty enough and easy to implement to have it in dextrose-2
14:55 m_anish alsroot, makes sense
14:55 anyway, re py, the schools here open on 22nd feb...
14:55 bernie silbe: for example: https://bugzilla.redhat.com/sh[…]bug.cgi?id=666928
14:56 m_anish initially we were planning on having a release on 1st feb, have it tested by the edu team/formadores/kids in 1-1.5 weeks... and we fixing the issues that crop up then...
14:56 bernie alsroot: would you like to work on reviving wadeb's logcollect patch then? it's here: http://bugs.sugarlabs.org/ticket/1439
14:56 alsroot m_anish: anyway, I'm waiting for you ping to discuss dates and TODO for dextrose-2 and prioritize tasks
14:57 satellit_ is now known as satellit_afk
14:57 silbe alsroot: FWIW, it's only those severe cases I'd be interested in in the first place. We already have enough other bugs to work on. And sadly, crashes actually still seem to happen sometimes for me (though I'm not sure exactly what is crashing and how, i.e. if it's even something abrt would catch). :-/
14:57 m_anish but it makes much more sense to have 2-3 iterations of test-release than one in terms of catching bugs (or have iterations in parallel)
14:58 alsroot silbe: for core dumps we can all time switch to more personal/detailed workflow
14:58 ..to work w/ bugs reporter
14:58 m_anish so we're actually making a release on this tuesday and have them tested out... meanwhile i'll get a list of priority task soon that we need to fix
14:59 tch m_anish: +1
14:59 bernie silbe: I think you're hitting the python gc bug. I have reported it a dozen times with abrt but nobody cares:
14:59 m_anish tch's notification, yum updater, microformat upd would be major additions to the build on top of current dx2
14:59 bernie https://bugzilla.redhat.com/bu[…]sification=Fedora
14:59 m_anish (plus 2 patches from tch )
14:59 alsroot bernie: yup, something like #1439 but maybe for dextrose it might be more implicit
15:00 tch m_anish: we also need to include my script to disable the mesh after sleep/week :)
15:00 m_anish I plan to keep a TODO-py page on wiki.sl.o
15:00 silbe has anyone tried pgf's kernel patch for disabling mesh altogether?
15:00 m_anish also, slightly unrelated, maybe we need to review http://wiki.sugarlabs.org/go/Dextrose/TODO
15:00 or http://wiki.sugarlabs.org/go/Dextrose/TODO-temp
15:01 alsroot m_anish: it act I was thinking about that when we will get more TODO items to work on
15:01 ..in fact
15:01 bernie m_anish: if there's no information loss, you can rename TODO-temp on top of the old TODO page
15:01 m_anish alsroot, yup
15:02 bernie, there's no info loss (only a repeat entry is removed)
15:02 bernie m_anish: ok then, let's replace the page. I think you did a great work, btw
15:03 m_anish bernie, will remove/replace, I guess that TODO is the dextrose(-3) roadmap?
15:03 oh, one more thing...
15:04 since we have the yum updater up we'll be having 2 (or three) repos : release, testing etc.
15:05 silbe BTW, we need to involve upstream early on for UI changes. We had too much wasted effort because upstream didn't agree with what Dextrose came up with.
15:05 m_anish and need to figure out a process to move rpm's from testing to repo, and align it with dx-mainline
15:05 s/testing to repo/testing to release/
15:05 bernie m_anish: yes... we could also rename TODO to Roadmap... but roadmaps need to have dates attached to them imho
15:06 m_anish: like alsroot's roadmap: http://wiki.sugarlabs.org/go/P[…]form_Team/Roadmap
15:06 m_anish: or our roadmap for dextrose 1: http://wiki.sugarlabs.org/go/Dextrose/Roadmap
15:06 tch silbe: if we can speed up the time required to get a consensus, sure.
15:08 silbe: the backup thing is a great example,  it took months just to not get a consensus..
15:08 silbe tch: our Design Team consists of volunteers with unrelated day jobs. So it's expected that it takes a fair amount of time. That's why it's so important to talk about UI changes as early as possible.
15:08 m_anish bernie, we could perhaps plan a roadmap (but our timeline is quite short, schools open 22 feb)
15:08 bernie silbe: until now, upstream was very dysfunctional and disconnected from the needs of deployments. there was no way dextrose or olpc could get something designed and implemented while working with upstream.
15:09 silbe: now, with you and erikos on the helm, I'm confident that something will change
15:09 tch silbe: yeah, i understand that, is just that sometimes deployments can't just sit and wait until consensus arrives, they will just do it..
15:09 bernie silbe: for UI design, I'm still very dissatisfied. look at my proposal for a super-trivial about box: it's been stuck for 2 weeks already.
15:09 tch silbe: and we need to take that into account if we want them to keep them together with mainstream
15:10 silbe bernie: re. the about box, consider that the past two weeks were "holidays", i.e. most people were busy with other stuff.
15:10 bernie silbe: making developers wait weeks on consensus is a productivity killer, we must stop doing this in sugar-devel.
15:11 m_anish what i'm hoping eventually is that we can use the testing branch to beta test new features (that _may_ not be requirements from the edu team) and keeping this branch shared across more than one deployment
15:11 bernie silbe: true. I just pinged the RFC to see if something moves
15:11 dfarning before we worry to much about design let's give silbe and simon time to prove that patch management is work
15:11 bernie m_anish: hopefully, yes.
15:11 alsroot bernie: "we must stop doing this", /me does excatly that all time :)
15:12 silbe bernie: in general, I think we'd be much better off if we discussed it early with upstream and _then_ implement it, instead of the other way round. We don't necessarily need perfect consensus, but we should at least agree on the general way before it's already implemented and too hard to change.
15:12 dfarning for the short time dextrose will differ from upstream.... we will start working with OLPC/SL to fill the UI designer gap.
15:12 bernie alsroot: :)
15:13 m_anish is done with is list of py topics :)
15:13 bernie silbe: I agree. in the past, starting UI design discussions failed all the time... now it may be different.
15:13 silbe dfarning: the problem with OLPC/SL taking over the design job is that we get inconsistent with each other and force upstream to take our stuff because users get used to it.
15:14 bernie dfarning: yes, that would help a lot.
15:15 silbe anyway, I don't expect everything to get perfect overnight. All I want for now is that we start UI design early on (way before a developer starts working on a task!) and try to get upstream involved, even if it doesn't end up reaching consensus.
15:15 bernie silbe: true
15:16 silbe: true^2
15:16 dfarning silbe, yes we still need to figure it out....  but as long as we are moving forward we can work out problems as we go.
15:16 silbe great! m_anish, sorry for interrupting again (it was at least triggered by the TODO list ;) ).
15:17 m_anish silbe, bernie could we elaborate more details on the TODO wiki page and have them run through sugar-devel (something like batch reviewing or UI proposals)?
15:17 s/or UI/for UI/
15:17 silbe m_anish: sounds good. let's review the list on IRC first and dextrose/sugar-devel later.
15:17 (but after the meeting ;) )
15:18 bernie (for the record, I think that the dextrose team always did involve upstream early, but failed to get consensus and ended up doing our own thing instead of blocking forever)
15:18 m_anish silbe, +1
15:18 bernie silbe: +1
15:18 m_anish bernie, true!
15:18 tch bernie: can't agree more :P haha
15:18 m_anish (do we have more topics to discuss?)
15:20 011
15:20 010
15:20 001
15:20 000
15:20 #endmeeting
15:20 meeting Meeting ended Sat Jan  8 15:20:20 2011 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot. (v 0.1.4)
15:20 Minutes: http://meeting.sugarlabs.org/s[…]-08T14:01:34.html
15:20 Log:     http://meeting.sugarlabs.org/s[…]11-01-08T14:01:34

Minutes | Index | Today     Channels | Search | Join

Powered by ilbot/Modified.