[09:21] Unity failed to build on arm. https://launchpadlibrarian.net/131652234/buildlog_ubuntu-raring-armhf.unity_6.12.0daily13.02.19-0ubuntu1_FAILEDTOBUILD.txt.gz [09:21] Seems to be "/build/buildd/unity-6.12.0daily13.02.19/plugins/unityshell/src/inputremover.cpp:426:73: error: cast from 'unsigned char*' to 'long unsigned int*' increases required alignment of target type [-Werror=cast-align]" [09:21] And on line 461 [09:21] sil2100, bregma ^ [09:21] Some new commit, hm? [09:22] Yeah, it built yesterday [09:26] mterry: let my try and fix that [10:16] mterry: I fixed the build failure probably, but a bit hackishly - trying to build the whole unity trunk now on ARM for testing [10:24] sil2100, thanks [10:35] sil2100: if you look at intel/ati results, we are back to 25/26 failures per config [10:35] waiting nvidia [10:38] sil2100: urgh, 38 failures on it :/ [10:39] mterry: ^ [10:48] huh? [10:49] didrocks: anthy tests for ibus failed as well this time, geh :| [10:49] sil2100: do you mind looking as well? [10:49] Since the weekend I have no indicators and no global menus anymore - maybe this is related ? [10:50] MCR_: using the daily release? [10:50] yes [10:50] yeah, was an issue on the ppa [10:50] but should be fixed by now [10:50] unfortunately not here... [10:51] MCR_: apt-cache policy ? [10:51] sil2100: is it what we are seeing in the video for nvidia? ^ [10:53] didrocks, 500 http://ppa.launchpad.net/unity-team/staging/ubuntu/ raring/main amd64 Packages [10:53] release v=13.04,o=LP-PPA-unity-team-staging,a=raring,n=raring,l=Unity staging ppa,c=main [10:53] origin ppa.launchpad.net [10:53] ? [10:53] MCR_: doesn't tell me the version :) [10:54] unity: [10:54] Installed: 6.12.0daily13.02.08bzr3162pkg0raring0 [10:54] Candidate: 6.12.0daily13.02.08bzr3162pkg0raring0 [10:57] MCR_: ah, you are using staging [10:57] MCR_: this is not the daily build ppa [10:58] didrocks, oh, there is a new PPA ? I did not know that. I'll search... [10:58] MCR_: you should just use raring [10:59] MCR_: it's getting the latest and greatest daily [10:59] which pass tests [10:59] ok, thx - I will purge the PPA then... [10:59] sounds like a good plan :) [11:00] :) === mmrazik is now known as mmrazik|lunch [11:18] didrocks, purged the PPA, now using Installed: 6.12.0daily13.02.08-0ubuntu1 - still no indicators and no global menus :( [11:30] MCR_: what version of libindicator do you have? [11:33] didrocks, libindicator7: [11:33] Installed: 12.10.2daily13.02.15-0ubuntu1 [11:33] (thanks 4 your help, btw) [11:33] MCR_: this version is from the daily-release ppa [11:34] isn't it? [11:36] didrocks - hmm, I purged the PPA, but maybe something went wrong with that - thanks for the help again - I'll investigate [11:39] ok :) [11:41] grrr === _salem is now known as salem_ [11:56] mterry: ah bugger, sorry about that, will fix [11:57] mterry: suprised we don't have ci jobs for unity on arm [11:57] sil2100: were you going to fix that or are you busy atm? [11:57] its easy I can fix it now [12:00] smspillaz: I'm testing my fix now, armhf package is still building though [12:00] On unity-team/ppa [12:01] sil2100: ah okay. Do you have a branch somewhere ? [12:01] smspillaz: https://code.launchpad.net/~sil2100/unity/fix_ftbfs_arm_alignment_hack [12:01] It's a hack [12:02] smspillaz, unity-on-arm ci was disabled due to speedness issues [12:02] I could have changed unsigned char to unsigned long as well and just cast it to unsigned char instead, but this one works too [12:02] bregma: isn't there an autolander job for it? [12:02] bregma: we have that for compiz [12:02] autolanding shouldn;t have to take up to half a day per commit [12:03] bregma: trunk also shouldn't get broken :) [12:03] if there is more or better hardware, it should get reenabled asap, that's up to QA [12:04] I suspect there are other, more urgent uses for arm builders right now [12:04] that's true [12:04] bregma: I wonder why we can't cross compile yet [12:04] cross building with dpkg is not really an option, it's still way too broken despite multi-arch starting to trickle in [12:04] sil2100: yikes, that is hacky, I wonder why reinterpret_cast from unsigned char * to unsigned long * doesn't work on arm [12:05] bregma: yeah I know, I've tried it myself it sucks [12:05] smspillaz: well, ARM and its misalignment issues does not allow casting from a smaller pointer type to a bigger one [12:06] The other-way around is fine though, also, void * is safe here [12:06] huh [12:06] That's hm, a really bitchy thing anyway [12:06] didrocks, just FYI: I solved the no-indicators problems now by adding the ppa:indicator-applet-developers/indicator-staging-ppa [12:07] Usually __aligned__ should be used to fix such problems, but here we're getting the array from X, so I didn't know if we could do it in this way [12:10] sil2100: did you see my ping/question about the videos on nvidia? [12:11] didrocks: ah, yes, I thought that I answered that: the videos show that something is wrong with detecting ibus enabled - since it seems as if anthy is enabled (I think?) but the test fails noticing - but it might be an AP issue, not a regression [12:13] sil2100: ok, do you want us to relaunch the tests to see if it's transiant? [12:14] (that doesn't prevent to continue looking at the old run) [12:14] didrocks: yes, please, this way we won't waste time [12:15] ok, let's do that o/o\o/ [12:15] I checked and nothing changed in trunk that could have broken ibus for one engine [12:15] (running) [12:16] sil2100: please, let us know when the armhf fix is merged so that we rebuild unity [12:34] didrocks: sorry about that one :( [12:35] smspillaz: no worry dude, I wished that we had those running as part of the merger :) [12:39] smspillaz: ok, unity built fine with that hack of mine - you had a nicer idea for fixing it, or can we try merging it in as it is ;) ? [12:39] sil2100: merge as is [12:40] smspillaz: if something better comes up, I'm sure we can re-do this - I'll just add a comment that it's a hack, so that no one starts getting all WTF? after a while ;) [12:40] sil2100: well, reinterpret_cast'ing from unsigned char * to unsigned long * is a hack too [12:41] it only works because libX11 and the server align the property data with the specified property type [12:43] smspillaz: can you approve it then :) ? Thanks! [12:44] sil2100: can you link me the mp? I don't get email notifications for unity [12:44] (cause I don't work for canonical anymore) [12:46] sil2100: done, you will need to change the global branch status yourself (I don't have permission to do that) [12:46] howdy kamstrup [12:58] smspillaz: \o/ [12:59] smspillaz, you do not need to work for Canonical to get those emails, should I subscribe you ? [13:00] MCR_: not sure if anyone would willingly like to recieve those ;p [13:00] ;) [13:00] just freaks like MCR ;) [13:00] smspillaz, https://code.launchpad.net/~unity-team/unity/trunk/+addsubscriber [13:07] smspillaz, a mnior fix: https://code.launchpad.net/~mc-return/compiz/compiz.merge-type-is-unsigned-fix/+merge/149267 [13:08] *minor [13:14] sil2100, got one for Unity as well: https://code.launchpad.net/~mc-return/unity/unity.merge-minor-fixes/+merge/149269 [13:17] MCR_: saw that, acked, thanks [13:18] smspillaz, it is *very* minor, but still ;) [13:20] sil2100, smspillaz: Thanx 4 the fast approvals :) [13:22] * smspillaz is still waiting on the builder to check the xorg-gtest stuff ... again :( [13:22] this time it failed because I merged something new in which didn't quite follow the dso linking rule [13:22] which the pbuilders are strict about === mmrazik|lunch is now known as mmrazik [13:25] smspillaz, urgh - btw, bug 1094650 is still valid here [13:25] bug 1094650 in compiz (Ubuntu) "Compilation of lp:compiz with gcc 4.8.0 fails with -WERROR because of warnings" [Medium,New] https://launchpad.net/bugs/1094650 [13:25] smspillaz, I have to turn off WERROR or compilation fails here... [13:27] MCR_: its in-progress on the gnome-end [13:27] ah, ok - thanks [13:27] MCR_: there is not much we can do about it [13:27] sure, it's not a big problem [13:28] MCR_: well, the problem is that its a non-standard compiler option, and passing that option to clang makes it error out [13:29] the error is about gcc though, but I've seen your report and their answer now... [13:36] MCR_: this is just why C++ can be a pain sometimes [13:42] smspillaz, btw - we have to party :) -> today MCR branch No. 100 will be merged \o/ [13:43] heh [13:44] MCR_: well, I will be 21 soon :) So maybe we can roll it into that [13:45] Sam, it is still incredible and a miracle for me, how someone can be such an C++ expert with this young age ;) [13:45] pfft, I know nothing [13:47] Congratulation to the 21 of course... Hopefully someone gets you a faster machine, so you do not have to fight with disk space and compiling times anymore... [13:48] MCR_: well, I'm waiting until machines with broadwell come out [13:48] of course, last year I told myself I was waiting until machines with IVB came out [13:48] ... and the year before sandybridge .... [13:49] The open source community should buy you such a machine each year ;) [13:49] heh === dandrader is now known as dandrader|afk [14:19] didrocks, the rm /var/crash hack didn't fix it [14:19] didrocks, but on the plus side, tests are under threshold now [14:20] didrocks, any objections if I make that percent->number change in the thresholds to their current values? (33 failed tests and 14 regressions allowed) [14:20] mterry: yeah, those are high enough values, aren't they above the threshold even? [14:21] didrocks, those are the threshold. I think 481 tests. 7% and 3% respectively [14:21] mterry: ok, perfect, I understood "current values" as, "current failure on nvidia" :) [14:22] didrocks, why do we allow so many regressions? darn tests should be reliable [14:22] mterry: well, I agree, sil2100 did his best to fix as many as possible, (remember that we went from 130+ failures to where we are now) [14:23] mterry: oh, btw, rebuilding unity alone now? :) [14:24] didrocks, I suppose... Maybe it won't crash this time. It looks... like maybe we crashed in between suites of autopilot (if the Date file inside the apport crash is reliable) [14:24] mterry: yeah, so a valid case for it [14:24] * mterry loads up his trusty jenkins web interface [14:24] tsss :p [14:24] * didrocks is so saddddd === dandrader|afk is now known as dandrader [14:31] didrocks, so the fact remains that compiz is crashing sometimes and we don't know why and can't get traces from it for some reason (why?) === dandrader is now known as dandrader|afk [15:08] sil2100: andyrock: Trevinho: https://code.launchpad.net/~compiz-team/compiz/compiz.fix_1110138/+merge/147548 free karma [15:11] smspillaz, in reviewing the debian/ changes to compiz, I noticed and filed bug 1130160 [15:11] bug 1130160 in compiz (Ubuntu) "Add copytex to default/unity profiles" [Undecided,New] https://launchpad.net/bugs/1130160 [15:11] mterry: ah, I knew I had missed it somewhere [15:11] mterry: thanks [15:12] smspillaz, really it was thanks to didrocks who was over my shoulder :) [15:12] mterry: are you running with the default settings ? [15:12] mterry: (+ upgrade) [15:13] smspillaz, um, I think so [15:14] mterry: can you post the output of [15:14] gsettings get org.compiz.core:/org/compiz/profiles/unity/core/ active-plugins [15:14] I haven't got unity installed at the moment and need to get the default plugin order [15:15] smspillaz, ['core', 'composite', 'opengl', 'compiztoolbox', 'decor', 'vpswitch', 'snap', 'mousepoll', 'resize', 'place', 'move', 'wall', 'grid', 'regex', 'imgpng', 'session', 'gnomecompat', 'animation', 'fade', 'unitymtgrabhandles', 'workarounds', 'scale', 'expo', 'ezoom', 'unityshell'] [15:15] hmm [15:16] copytex isn't there [15:16] mterry: cat ~/.config/compiz-1/compizconfig/done_upgrades ? [15:17] smspillaz: could that be the reason for why popey's decorator was crashing when using the patch you prepared for VB issues? [15:17] smspillaz, empty... maybe I'm not on the latest daily-build version yet [15:17] sil2100: yeah probably [15:18] smspillaz, I have the latest compiz from daily-build [15:19] mterry: just to double check, ls /etc/compizconfig/upgrades ? [15:19] hmm wait [15:19] that doesn't seem right [15:19] smspillaz, 01, 02, and 03 [15:20] hmm [15:21] I think I changed the install dir to $datadir/compizconfig/upgrades on didrocks request but the .install file didn't get updated [15:23] mterry: okay, branch coming your way [15:24] mterry: https://code.launchpad.net/~compiz-team/compiz/compiz.fix_1130160/+merge/149308 [15:29] smspillaz, thanks [15:35] smspillaz, sure [15:35] andyrock: cheers [15:36] smspillaz, quick question [15:37] ing *Window::glPaint (during occlusion test) windows are passed from top to bottom right? [15:40] andyrock: yep [15:42] here the first passed window is nautilus [15:43] smspillaz, maybe the desktop window? [15:43] andyrock: yep [15:43] andyrock: what are you trying to do ? [15:44] smspillaz, fixing bug 748539 [15:44] bug 748539 in unity (Ubuntu) "Multi-monitor - Panel and launcher visible on top of multimonitor non-focused fullscreen apps" [High,In progress] https://launchpad.net/bugs/748539 [15:45] just debugging for the moment [15:52] andyrock: heh, have fun with that one [15:53] andyrock: I'm pretty sure panels and such are meant to go above unfocused fullscreen applications [15:53] thats how compiz' stacking policy works [15:54] though maybe only if there's a window that needs to go above that fullscreen window [15:54] ... which is how it works if there's another one on another monitor [15:55] eg normal window > docks > fullscreens > other docks > other normals [15:55] maybe we could make it so that this policy only applies where they are on the same monitor, though that feels tricky [15:58] we just need to don't draw unity on monitor A if the first window (in the stack) on monitor A is a fullscreen one [15:58] maybe there are corner case [15:58] i'm testing this solution now [15:59] of course during alt+tab unity should be drawn [15:59] andyrock: that's true, but "don't draw unity on monitor B" is not really a concept that nux understands [16:00] maybe you can get away with it in the multimonitor case because we have to do clipping [16:00] but selectively painting BaseWindow objects ... that's something I spent about a month hacking into nux === dandrader|afk is now known as dandrader [16:38] bschaefer: if you have time for reviews, feel free to look at these two: [16:38] erm === francisco is now known as Guest89363 [17:30] smspillaz, the latest compiz autolanding run seems to have hit a merge conflict with debian/changelog entries. Note that if your branch is linked with a bug, you don't need to edit the changelog at all. I'll fix the merge conflict, just heads up [18:02] didrocks, https://code.launchpad.net/~mterry/compiz/latestsnapshot/+merge/149357 === salem_ is now known as _salem