/srv/irclogs.ubuntu.com/2011/06/01/#ubuntu-arm.txt

=== prpplague is now known as prpplague^3
=== prpplague^3 is now known as prpplague
=== prpplague is now known as prpplague^3
=== prpplague^2 is now known as prpplague
=== asac_ is now known as asac
=== lilstevie|ZNC is now known as lilstevie
=== txwikinger2 is now known as txwikinger
=== lilstevie is now known as lilstevie|ZNC
=== lilstevie|ZNC is now known as lilstevie
nemozHello, I would like to know what wifi usb devices people have used successfully on the beagleboard-xm with Ubuntu?11:31
nemozARM Cortex-A811:31
ogra_plenty11:32
tonuyes, plenty11:32
tonuit does not matter is beagleboard or not. Anything supported on linux works11:32
ogra_right11:32
ogra_well, at least anything supported on ubuntu :)11:33
ogra_minus ndiswrapper cards indeed :)11:33
nemozSo it should be pretty safe buying any of the usual wifi dongles?11:33
ogra_right11:33
ogra_you will have the same issues you have on other arches (if you have issues)11:34
tonuI used ALFA ones with RT8187 chipset11:34
nemozI was reading things about wifi devices not being recognised correctly by the device so I thought I would make certain11:34
tonualso some others but ALFA is something I always can recommend11:35
hrwI use tplink with ath9k_htc module11:40
nemozJust plug and go? Or any extra hoops11:46
hrwjust11:46
nemozactually there is an extra constraint, are the devices capable of creating and joining an ad-hoc network?11:54
* ogra_ never tried adhoc on omap11:55
hrwI have AP under desk11:56
zulogra_: ping13:47
ogra_zul, yeppers13:47
zulogra_: hi, so im suppose to press control-esc when the machine is powering on with the usb and the card plugged in right?13:48
ogra_no card needed13:49
ogra_but i wont do harm either i guess13:50
zulok so why is the usb needed again?13:50
ogra_the LEDs should light up, the screen should stay black13:50
ogra_the usb is the connection you flash the boot partitions on the internal MMC with13:50
ogra_from the host computer using nvflash13:51
* ogra_ needs to reboot, brb13:51
zulok..13:51
ogra_re13:55
zulogra_:  hey and where do you extract the linux4tegra tarball to?13:57
ogra_your PC the USb is connected to13:57
zuljust anywhere on that pc?13:58
ogra_yes14:00
loologra_: Hey, while updating the list of images in a local mirror, some questions popped up; is there still a wiki page with the daily images which you folks care about?14:01
loolI've noticed that there is an ubuntu-netbook project where the dailies aren't under ports: http://cdimage.ubuntu.com/ubuntu-netbook/ also I thought netbook was replaced by regular "ubuntu" (desktop)14:01
ogra_it wasnt replaced yet, thats A214:02
loolso that image is going away, right?14:02
ogra_there should be a release manifest wikipage page, not sure its there for oneiric yet14:02
ogra_yes, ubuntu-netbook-preinstalled will just become ubuntu-preinstalled14:03
zulogra_: ok duh...i got it now14:13
=== prpplague is now known as prpplague^2
zulhow long is flashing suppose to take usually?14:56
ogra_a few seconds14:59
ppisatiguys, what's the status of PM on omap4?15:12
zulogra_:  ok its working now thanks for the help15:13
=== brendand_ is now known as brendand
brendandhi, i'm trying to boot an omap4 natty image on a pandaboard15:25
brendandmy SD card has 2 partitions, i 74MB and 1 2.6GB15:26
StevenK~,.15:41
StevenK~>15:41
=== zyga is now known as zyga-afk
=== apachelogger is now known as transitlogger
=== jkridner_ is now known as jkridner
ogra_Daviey, http://people.canonical.com/~ogra/tegra/2.6.37/README first shot of the installer ... should make your life easier22:48
ogra_NCommander, ^^ in case you want to play ^^^^22:53
Davieyogra_: groovy.. what boot img should i use?22:53
ogra_http://people.canonical.com/~ogra/tegra/2.6.37/22:53
ogra_download the files there22:53
zulsweeeet22:53
ogra_if it works it should make life easier ... i only tested it locally yet22:54
ogra_(the ac100 models all differ a bit)22:54
ogra_persia, in case yu are still intrested in setting up a wikipage for the ac100, http://people.canonical.com/~ogra/tegra/2.6.37/README should be a first start22:57
Davieystandard platform++22:57
ogra_heh22:57
persiaogra_, Nice improvement.  It's worth mentioning that the 8GB model should have stuff on /dev/mmcblk0p7 and the 16GB model on /dev/mmcblk0p12 (and one needs to appropriately adjust the configuration).  I'm mostly waiting for you to confirm that it no longer requires a PPA and has a real image.23:01
ogra_persia, no, the installer hides all that details23:01
persiaI'm not sure it's a good idea to have things automatically install to the "first partition", as this may not have the right amount of space.23:02
ogra_??23:02
persiaOh, nevermind.  I read README incorrectly.23:02
persiaIt takes the image *from* the first partition.23:02
persiaHow are you detecting the appropriate target partition?  Just finding the largest?23:03
ogra_yeah23:03
ogra_yep23:03
* ogra_ will package all that stuff for oneiric, natty is my testbed atm23:03
ogra_then you can look at the code :)23:03
* persia still thinks that the regular installers should be used, and that the speed issue should be solved generally, but this is a separate issue.23:03
* persia is extra uncomfortable about having unreviewable installation code, and hugs the old version of the SD image that allows one to debootstrap23:04
ogra_well, i'm just to tired to make a branch and upload atm23:05
persiaNow is not a good time for you to upload :)23:06
ogra_its just three initramfs files23:06
ogra_and one script that adds an overlay to the initrd23:06
ScottKrsalveti: Looking at Bug #791250 which you filed (I believe with the aid of a script).  It says "This log snippet might be of interest, since it triggered the matcher 'Purging chroot-autobuild'".  Wouldn't all build failures trigger on that?23:06
ubot2Launchpad bug 791250 in kdegames "kdegames version 4:4.6.3-1ubuntu1 failed to build on armel" [High,Triaged] https://launchpad.net/bugs/79125023:06
persiaI don't really like that fix: there are several GL-capable video devices available for armel devices (note that this class of fix is pervasive, unfortunately, so this specific example isn't in any more need of fixing than anything else)23:08
rsalvetiScottK: there can also be "Source-dependencies not satisfied" for broken dependencies23:08
rsalvetiand yes, it's a script23:09
rsalvetilp:svammel23:09
ScottKTrue.23:09
ScottKI guess it seems like a pretty broad brush.23:09
ScottKWe don't really need a bug for every build failure.23:09
ScottKWhen I'm curious about build failures I look at http://qa.ubuntuwire.com/ftbfs/ and not the bug tracker.23:10
ogra_++23:10
rsalvetiScottK: we're tracking build failures with bug because the it's easier to go over the bug list at launchpad23:10
rsalvetiand use that at the arm porting queue23:10
rsalvetiand also put more folks involved on this23:11
persiaIn the case where the build failure is particularly interesting, and someone has an idea to address the class of problem, bugs can be nice, but that's a special case.23:11
ScottKrsalveti: Easier for who?  It's more stuff to mess with for us.23:11
persiarsalveti, Please don't.  It annoys other developers.23:11
rsalvetithe script can also close the bugs once a newer package version is uploaded without build failures23:11
ScottKSeems like pointless paperwork.23:11
ogra_super pointelss paperwork23:11
ogra_and annoying too23:11
rsalvetiI don't think it's pointeless23:11
ogra_since now everyone is forced into that paperwork ... else we end up with tons of open bugs23:12
rsalvetiit helped identifying easily the issues we had with binutils and ld23:12
ogra_the build logs didnt suffice ?23:12
rsalvetiand I really prefer a bug to be filled than just let it at the ftbfs list23:12
rsalvetiwell, you need to go over all logs23:12
rsalvetibugs you can trace status23:12
persiarsalveti, The issue is that FTBFS is transient.  We did an automated build-failure-bug-report thing before (back in Dapper), and stopped because by the time people looked at many of the bugs, they had been addressed by some other upload addressing the transient issue.23:12
rsalvetiwho is taking care of that23:12
rsalvetidetails about what happened and such23:13
persiaConsidering the number of packages that get updated for reasons having nothing to do with build failures, often in ways that affect them (new upstream versions, Debian syncs, etc.), it's especially pointless this early in the cycle.23:13
rsalvetipersia: that's why the script can also close bugs23:13
ScottKrsalveti: I think it may be different near release, but in this stage of the cycle, it's just a lot of churn.23:13
ScottKrsalveti: All of which doesn't keep people's inbox from getting clogged.23:13
persiarsalveti, Why not just add status tracking to the FTBFS page, if that's what you need.23:13
rsalvetipersia: why not bugs?23:14
rsalvetiwhat's the problem with this?23:14
rsalvetinoise?23:14
ogra_yes23:14
rsalveticome on, it's a good thing to track the issues there23:14
ScottKLow signal to noise ration.23:14
ScottKrsalveti: We disagree.23:14
ogra_noise, mail, harder to use interface, slowness23:14
rsalvetiif we're maintaining the bug list23:14
persiaNot just noise, but noise that often has zero value.  Noise with value can be tolerated.23:14
rsalvetiif you point me that the script is really creating just noise and no value, then it's ok23:15
ogra_but most annoying as i said above, i need to invest extra time into weeding through LP if i gave back a package23:15
rsalvetifor now it's just people complaining because of noise23:15
ScottKrsalveti: If it files a bug for a random FTBFS, it's noise.23:15
ogra_(or even fixed the buildfailure)23:15
persiarsalveti, We're so pointing.  Can you demonstrate some value?  Do you need anything other than status?23:15
rsalvetipersia: just check the bugs that were filled and changed status23:16
persiaIt would be fairly easy to have the FTBFS tracker export some JSON, so you could parse it more easily (not that it's that hard to parse in the current state)23:16
rsalvetifor example, there are a lot of bugs for the gles porting for Qt23:16
persiarsalveti, So, what value does this bring over the existing FTBFS tracker?23:16
Davieyogra_: Ah nice!  It works directly from the tarball23:16
rsalvetiit's a good thing to have bugs for that23:16
ogra_Daviey, does it work for you ?23:16
rsalvetipersia: it's just a tracking tool23:16
rsalvetithat's already there23:16
rsalvetiand we can easily use23:16
rsalvetiif you want to improve the ftbfs to track the status I'd be fine with it23:17
persiarsalveti, I disagree, but that's because I think that the assumption that armel == GLES is fundamentally flawed, and have hardware that you are breaking with the "transition"23:17
Davieyogra_: Don't know yet... i copied the contents of the tgz and got "No tarball found".... so now putting the tarball in place :)23:17
ogra_rsalveti, if we can get around drowning in pointless bugs thorough that23:17
ScottKrsalveti: Most of the tasks on the Qt GLES bug were added manually.  That one is a bit of an exception.23:17
ScottKThat was a case of trying to coordinate effecting deliberate change.23:18
ScottKThese auto bugs are completely different.23:18
ogra_Daviey, heh, damned, READMEs dont allow <blink> :)23:18
rsalvetiwell, I guess I can only put some more manual work before actually filling the bugs23:18
rsalvetitrying to identify it's valuable or not23:18
persiaogra_, You could have written it in HTML: you are asking people to pull over HTTP after all...23:18
ScottKrsalveti: Do you think filing a bug for every compiler warning in a build log is a good idea?23:18
ogra_persia, yes, as well as i could have uploaded my code to a branch :P23:19
Davieyogra_: heh, i am a typical man... try first, read the instructions afterwards :(23:19
persiarsalveti, That would have *huge* value, if you did.  Bugs with notes about *why* something FTBFS, or other useful information are very useful.23:19
ScottKThis is basically the same kind of noise only a lesser scale.23:19
rsalvetiso what you're saying is that is basically useless to track ftbfs in bugs in general?23:19
ScottKrsalveti: I agree with persia.  If the bug includes some human pre-processing to indicate what went wrong or the class of fix, then it could have value.23:20
rsalvetiok, that's something we can improve23:20
persiaNo.  It's useless to have automatically-filed bugs with no additional information.23:20
ScottKrsalveti: It's useless to track them in the bug tracker.  We have other tools for htat.23:20
Davieypackages in Main, or ones a particualr team are going to work on for certain has value in having bugs for FTBFS IMO.23:20
ogra_rsalveti, i usually open a bug if a first poke on a package didnt help, but i would never do it automatically23:20
ScottKDaviey: I disagree.23:21
persiaDaviey, How do components affect this?  I don't see any relation.23:21
rsalvetiI also don't like the idea of just going over and filling without any kind of filter23:21
rsalvetithat's why I believe that if this is fixed, it'll actually bring value23:21
rsalvetiand not just noise23:21
ScottKUnless there's more information available than what can be found in the build log, it doesn't help.23:21
DavieyScottK: I know you disagree.. we'll have to agree to differ :)23:21
Davieypersia: In so far as Canonical /has/ to care about supporting Main, and if a particualr team is committed to certain packages it helps tracking.23:22
persiarsalveti, What sort of filter do you imagine?  For me, the threshold is a filter that requires human intelligence: if we can figure it out in an automated manner, I'd rather have a realtime reporting system.23:22
rsalvetipersia: manual work before actually filling it23:22
rsalvetichecking at least if it looks like a bug23:22
Davieyogra_: 30 mins! bah.23:22
rsalveticurrently the script just goes and fill everything23:23
rsalvetidoesn't even ask if you want to check it first23:23
ogra_Daviey, really depends on your devices ... 30 is worst case23:23
persiarsalveti, If you did that, I'd be arguing that you were doing a good thing by filing the bugs :)23:23
rsalvetisure, that's something I can do23:23
persiaCool!23:23
Davieypersia: A random package in universe that has never been touched in Ubuntu.. where it is most unlikely that any ubuntu contributor is going to fix, would just be a noise raising a bug.23:24
rsalvetijust tried today as slangasek was doing for past cycle23:24
persiaDaviey, And how does this differ from a random package in main that hasn't been re-uploaded since the Debian import in Warty?  I agree with the thoughts behind your statement, just don't believe it has any relation to components.23:24
ScottKrsalveti: And if (thinking down the road) you had tags for classes of FTBFS bugs then that would be super useful as if someone has fixed (as an example) a linker failure in a cmake based package, fixing the second one of those is pretty easy.23:24
DavieyIf we rely on using a FTBFS tracker for packages we really care about... we need somewhere to link branches, diffs and discussion.. so the FTBFS tracker needs enriching. Would also be nice to be able to assign it to someone.. And track as Work Items.  Suddenly, the FTBFS tracker has re-implemented launchpad. :)23:25
persia(nor that it is specific to Canonical: same applies for any of the other firms that sponsor developers to do stuff)23:25
persiaDaviey, Talk to the LP team: there is work *within* LP to integrate the FTBFS tracker.  That doesn't mean we should be using bugs.23:25
rsalvetiScottK: yup, also thought about that23:25
ScottKDaviey: I typically don't do any of those things with FTBFS, I just fix stuff.23:25
rsalvetihaving specific tags can for sure help a lot23:25
Davieypersia: I agree.. the fact is, I have raised FTBFS bugs for packages the server team care about.. Other teams can, or cannot do the same if they like. :)23:26
ScottKrsalveti: You might also talk to dholbach about integrating that thought with the work he's doing on helping getting new developers involved.23:26
DavieyScottK: Well tracking work that needs to happen in the cycle helps to be able to check we are getting stuff done, and not duplicating effort.23:26
persiaDaviey, Implicit in that action is the presumption that nobody external to that team is likely to do anything.  I think such a presumption is actively harmful.23:26
Davieypersia: hmm, no23:27
rsalvetiScottK: yeah, will talk with him23:27
Davieypersia: you are missundertanding me23:27
ScottKDaviey: you don't need a bug to have a branch, so that's off your list.23:27
ScottKWork items are for features, not bug fixing.  So that's out.23:27
DavieyHaving a bug against a package is being MORE open so other teams, within or external to Canonical can see the progress of the issue being resolved.23:27
ScottKSo except for assignment and discussion, I think the FTBFS page is fine and what fraction of FTBFS fixes need any of that?23:28
DavieyIf it's just remembered in my or ScottK's head... how can others see that it is being noticed and worked on?23:28
Davieydirected to persia ^^23:28
ogra_Daviey, but we have a list where it shows up already23:28
ogra_why do we need another one ?23:28
Davieyogra_: I haz ubiquity!23:28
persiaDaviey, That's why we have an FTBFS tracker, and why the LP team is working to improve the available overviews of FTBFS status *within* launchpad.  It needs to be available, but having bugs is needlessly duplicative, unless there is useful value added in the bug.23:28
ogra_it even categorizes packages by tasks23:28
ogra_Daviey, awesome !23:29
ScottKDaviey: Based on your logic, instead of MoM we should have it autofile bugs.23:29
persiaWell, by packagesets, so that groups (like the server developers) can easily see which packages interest them.23:29
Davieypersia: but tracking work that needs to happen in the sub-cycle makes it easier aswell.23:29
persiaScottK, We used to do that too :)23:29
ScottKpersia: I know.23:29
persiaDaviey, Yes, but the defect management system is the wrong way to "track work".23:29
DavieyI do /not/ see the harm it has in having tracking bugs via this method util we have something better.23:29
Davieypersia: FTBFS is a defect, no?23:30
ScottKWe already have something better.23:30
persiaNo, because it's transient: see above.23:30
Davieywe do not.23:30
DavieyAs i said at the start of this... i'm not trying to influence you to change what you think is better or not.23:31
persiaIf you need WIs, the FTBFS tracker should be extended to allow folks to have stuff show up as WIs.  This is unrelated to polluting the bug tracker.23:31
ScottKDaviey: The only thing we lack that a bug gives is the ability to assign someone work.  If someone is going to assign someone to fix a FTBFS then in that rare instance said someone can file a bug.23:31
Davieylast cycle, when the server team tracked FTBFS issues via bugs - we got MORE community contributions than we had before we had the bugs.23:31
ScottKThat's far more efficient than N developers subscribed to bugs dealing with stuff in their inbox.23:31
persiaAnd assignment inherently assures that nobody other than the assignee will do the work.23:31
DavieyWhy do you both feel so strongly about this?23:31
persiaDaviey, Is this because of the bugs, or because of a lack of socialising potential server contributors to the existing tools?23:32
ScottKBecause I get so much bugmail already that I really dislike seeing even more of it autogenerated and thrown into my inbox for no point.23:32
Davieypersia: Well probably both...23:32
persiaBecause we did it before, and we drowned, and we turned off the automated systems, and it's *really* frustrating to encounter the same issue again, with certain forknowledge that it fails to scale.23:32
DavieyScottK: Well thanks to LP's latest bug mail filtering, that should be easier.23:33
NCommanderDaviey: LP's bug filtering is still very limited23:33
ScottKDaviey: Not really.  I still don't seem to be able to unsubscribe from bugs as an individual without unsubscribing the entire team.23:33
NCommander+1 ScottK23:33
persiaDaviey, No, because that makes everyone else do work.  Is the total sum of that work worth the benefit?  How does it compare to the total volume of work required to extend the existing tracker to meet your needs?23:33
ScottKSo the improvements turn out not to help much at all with the general use case I have.23:34
DavieyOkay.. The only complaint i had last time was from ScottK.  If there were a handful of complaints, then i'd be a little more understanding.23:34
ScottKDaviey: I think persia is referring to something general.23:35
NCommanderDaviey: for FTBFS that require cooridination or input, bugs are opened and generally go a long way. For the five minute fix type, not so much23:35
DavieyWell the fact remains, it got results last time.23:35
persiaI am.  Specifics are relatively uninteresting to me.  I actively avoid any mail originating from launchpad anyway.23:35
micahgScottK: unsubscribing as an individual I think was just fixed23:35
persiaDaviey, Sure, but I stand by my argument that this is due to not exposing the common resources used by Ubuntu Developers to nascent members of your team combined with a poor definition of the current "server" packageset, making the current report insufficiently interesting to your team.  Both of these need be sorted *anyway*.23:36
Davieyagreed23:37
persiaDaviey, And, please, if you do need WI tracking for FTBFS, at least file a bug against the FTBFS tracker and the WI tracker (two tasks).  You may or may not have time to actually enable these features, but the above IRC log is not a good way to capture the requirements (and the primary developers of both of those tools are not in this channel)23:40
Davieypersia: ok23:45
DavieyScottK: Do you filter your bug mail on X-Launchpad-Bug-Tags ?23:46
persiaThank!23:46
ScottKmicahg: Just as in when?  I tried a couple of days ago and couldn't.23:46
ScottKDaviey: No.  I have a mailbox that it all goes into.23:46
micahgScottK: well, I'm in the beta test group, so I think I've had the feature for a couple weeks23:47
DavieyScottK: crikey.23:47
persiamicahg, Wasn't that allowing individuals to unsubscribe from specific bugs where they are subscribed to the package?  This is different from being a member of a team that is subscribed to a bug.23:47
micahgScottK: http://blog.launchpad.net/bug-tracking/coming-soon-improved-bug-subscriptions, it's coming soon :)23:48
micahgpersia: as I understood it, it was both23:48
persiaHrm.  I hope so.23:48
ScottKmicahg: When I tried it a couple of days ago I could unsubscribe the team from the bug, but not just myself.23:49
ScottKDaviey: I've experimented with more detailed sorting, but find all in one mailbox and use local search tools is more efficient for me than trying to predefine how I want stuff sorted out.23:50
micahgpersia: ScottK: right, so it's just the beta team ATM, it's the mute bug feature that does it, it's not subscription based23:52
ScottKAh.  OK.23:52
ScottKThere's a lot of release team bugmail I'll stop seeing soon after that lands then.23:52
micahgit covers both use cases mentioned (subscribed to package/product and team mail that's not going to a ML)23:52

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!