smspillazslangasek: you'll need to talk to the people from distro about it02:10
=== mmrazik is now known as mmrazik|afk
didrockshey duflu!06:08
dufludidrocks: Good morning06:10
didrocksduflu: FYI, the 0.9.9 wasn't that raring compatible, you missed 2 new features that were in and not in raring + a string break06:12
didrocksduflu: has to finish really late to convince the release team to get it in on Friday :/06:12
dufludidrocks: Hmm. The diff said I had a clean copy of lp:compiz/raring :/ ...06:13
dufluOK, please point out what's missing and I can fix it06:13
didrocksduflu: well, it's in now, but from: https://lists.ubuntu.com/archives/raring-changes/2013-April/008744.html06:14
didrocksduflu: there are:06:14
didrocks  * Showdesktop plugin: Wishlist/Feature-Request: Implement "Random"06:14
didrocks    movement direction option (LP: #1161343)06:14
ubot5Launchpad bug 1161343 in Compiz "Showdesktop plugin: Wishlist/Feature-Request: Implement "Random" movement direction option" [Undecided,Fix committed] https://launchpad.net/bugs/116134306:14
didrocksand   * [needs-packaging] Wishlist: Missing plug-In: Freewins (Freely06:15
didrocks    Transformable Windows) (LP: #1012194)06:15
ubot5Launchpad bug 1012194 in compiz (Ubuntu) "[needs-packaging] Wishlist: Missing plug-In: Freewins (Freely Transformable Windows) -> not shipped in any .install file in raring" [Wishlist,Fix released] https://launchpad.net/bugs/101219406:15
didrocks(the second is a false listing I guess)06:15
didrocksduflu: also, there is in those merge Show desktop -> Show Desktop06:15
didrocksduflu: I think this is breaking translation, so should be reverted06:15
dufludidrocks: I'm confused. That stuff was never in lp:compiz/raring ...06:16
didrocksduflu: no, it was in 0.9.906:16
didrocksduflu: we landed /raring in raring06:16
dufluI didn't go missing. It just moved to lp:compiz/0.9.1006:16
didrocksand you wanted us to land 0.9.906:16
didrocksso, the content of 0.9.9 needed to be raring compatible06:16
didrockswhat I asked and you told me yes before I moved to that branch06:17
slangasekdidrocks: hi, wrt bug #763148, how do you want this merge proposal for 0.9.9? Smash everything into a single cherry-pick commit?06:20
ubot5bug 763148 in compiz (Ubuntu) "Adding/Removing an external monitor causes open windows to move to another workspace" [Medium,In progress] https://launchpad.net/bugs/76314806:20
didrocksslangasek: hey, still awake? ;) yeah, a single commit is fine, you can just reference the other one in lp:compiz if you wish in the comment06:21
slangasekdidrocks: ok, cool06:24
mzanettiSaviq: good morning07:36
Saviqmzanetti, hey07:36
mzanettiSaviq: https://code.launchpad.net/~mzanetti/unity/more-stats/+merge/15744107:36
tsdgeosmzanetti: why this07:48
tsdgeos-if(buildtype_lower MATCHES coverage)07:48
tsdgeos+IF(CMAKE_BUILD_TYPE MATCHES [cC][oO][vV][eE][rR][aA][gG][eE])07:48
Saviqtsdgeos, 'cause we're lowering the build type?07:48
tsdgeosSaviq: read the +/.07:48
mzanettiSaviq: that was working previously too07:48
Saviqtsdgeos, before it matched cOvErAgE, too07:49
mzanettitsdgeos: because all other projects use it this way and our jenkins hooks use this string to identify if a project supports coverage07:49
Saviqtsdgeos, I can't see nothing wrong with that?07:49
tsdgeosit's making something more complex for no reason?07:49
mzanettitsdgeos: you think I should change the jenkins hook instead? that would be possible too. I opted for this for consitency with apps07:50
Saviqmzanetti, no07:50
tsdgeoswhat's wrong with "if(buildtype_lower MATCHES coverage)"?07:50
Saviqmzanetti, that will work ^07:50
* Saviq has eyes problems07:50
mzanettiare you guys getting my messages?07:50
Saviqmzanetti, yes07:50
tsdgeoswe are07:50
Saviqmzanetti, we have the "buildtype_lower" var07:50
Saviqmzanetti, that's CMAKE_BUILD_TYPE.toLower()07:50
mzanettiSaviq: yeah... as I said... I either revert this change and change our jenkins hooks07:51
Saviqmzanetti, and the empty endif() is "new style"07:51
=== alan_g is now known as alan_g|afk
Saviqmzanetti, why would you change it?07:51
Saviqmzanetti, there's no need to change?07:51
mzanettibecause all other projects use it this way and our jenkins hooks07:51
tsdgeoscOvErAgE is properly lowercased to coverage, no?07:51
mzanetticopy paste not working either07:51
Saviqmzanetti, or do you just mean for the sake of consistency?07:52
mzanettiSaviq: ^07:52
mzanettiSaviq: I can revert the change here and change the jenkins hooks07:52
Saviqmzanetti, then yeah, the "new way" is to lowercase it07:52
Saviqand match the exact string07:52
tsdgeossorry... :D07:54
Saviqmzanetti, tsdgeos, does make testIndicatorRow work for you?07:58
* mzanetti checks07:58
tsdgeosit did on friday07:58
Saviqtsdgeos, but it's using ChewieUI directly07:59
tsdgeosand it doesn't now07:59
* tsdgeos scratches head07:59
mzanettino... doesn't work08:00
mzanettinote for myself: we need to enable the qmluitests in CI asap...08:01
tsdgeosyes we do :D08:02
tsdgeosSaviq: it's not using chewieui08:04
Saviqdednick, actually, wanted to ask here08:04
Saviqtsdgeos, yeah, I see08:04
tsdgeosit's using a fake one08:04
tsdgeosmaybe the problem it's not finding it08:04
Saviqdednick, do the IndicatorRow tests work for you?08:04
tsdgeosahh, i think i see the problem08:07
tsdgeosit has to do with the move of builddir != srcdir08:08
Saviqdednick, yes08:08
Saviqdednick, ^08:08
tsdgeosi'll propose a MR08:08
Saviqtsdgeos, cheers08:08
Saviqdednick, tsdgeos is on it08:09
=== alan_g|afk is now known as alan_g
dednickif you're testing using qmltestrunner, you need to include the test dir import path. But it should be working using make qmluitests.08:10
tsdgeosSaviq: dednick: https://code.launchpad.net/~aacid/unity/make_indicator_row_test_work_again/+merge/15760008:13
mzanettiSaviq: tsdgeos: updated https://code.launchpad.net/~mzanetti/unity/more-stats/+merge/15744108:13
mzanettiSaviq: tsdgeos: requires this now too: https://code.launchpad.net/~mzanetti/pbuilderjenkins/update-some-hook/+merge/15759908:13
Saviqmzanetti, hmm, the grep won't match, will it?08:14
dednicktsdgeos: i think the hud may need the same?08:15
tsdgeosdednick: nope08:15
mzanettiSaviq: I changed buildtype_lower to cmake_build_type_lower to at least allow a little bit of consistency08:15
tsdgeosdednick: hud is actually compiling things08:15
Saviqmzanetti, yeah, I see08:15
Saviqmzanetti, nice hack, btw :D08:15
mzanettiSaviq: for what?08:15
Saviqmzanetti, the hook08:15
Saviqmzanetti, the grep through CMakeLists.txt ;)08:16
mzanettiSaviq: that was here before my :) I think mmrazik created those initially08:16
Saviqmzanetti, is there autolanding for lp:pbuilderjenkins ?08:16
Saviqmzanetti, yeah, I'm not saying it's yours, don't you worry ;)08:17
mzanettiSaviq: autolanding yes. but I need to manually upgrade the package on all jenkins nodes08:17
Saviqmzanetti, but that's why we didn't understand the need to update the hook ;)08:17
mzanettiSaviq: hehe, yeah08:17
dednickSaviq, tsdgeos: ah, my cmake files were old. hadn't updated to use builddir. wihch is why it was working for me. All fixed.08:20
dednickdoes anyone know why when i load an item in the testrunner it's smaller thatn the size set? loading with qmlscene sets the correct size.08:25
Saviqdednick, did you set the size on the top level component?08:30
dednickSaviq: yes08:30
Saviqdednick, hum, not sure, qmltestrunner is just a QQuickView as any other...08:31
dednickSaviq: seems to happen with all the qmluitests08:35
mzanettidednick: Saviq: yep... known bug08:35
mzanettidednick: should be fixed in Qt 5.108:36
mzanettidednick: shouldn't cause issues with tests though...08:36
mzanettiSaviq: there's no reason you didn't top-approve except the order of merging, right?08:37
Saviqmzanetti, I usually wait for Jenkins to do its thing08:37
dednickmzanetti: ok. thanks.08:37
mzanettiSaviq: ack08:37
mzanettiwhich is now... all fine... I'll top-approve08:38
Saviqmzanetti, beat me to it :P08:39
mzanettiSaviq: sorry... I forget that every time... may I ask you to approve this as mmrazik is ooo today? https://code.launchpad.net/~mzanetti/pbuilderjenkins/update-some-hook/+merge/15760108:41
Saviqmzanetti, sure08:42
tsdgeosmzanetti: just wondering, what's missing for the qmluitests integration thingie?08:44
mzanettitsdgeos: the MP with the coverage thingie from before08:44
tsdgeosso almost there?08:45
mzanettitsdgeos: yep08:45
tsdgeoscool :)08:45
Saviqmzanetti, approved08:49
tsdgeosCimi: ping08:51
Cimitsdgeos, pong08:51
tsdgeosCimi: if running tst_ResponsiveFlowView.qml with qmlscene can you open the comboboxes?08:51
Cimitsdgeos, mmm not if Iremember08:52
tsdgeosmay be that it used to work for the other test?08:52
tsdgeosi do remember playing with that manually08:53
tsdgeoswas i dreaming08:53
Cimitsdgeos, try responsivegridview08:53
tsdgeosdoesn't work either08:53
tsdgeosmaybe the ui toolkit changed?¿08:54
Cimitsdgeos, asksdkguys?08:55
tsdgeoswhat i am asking here is08:55
tsdgeosit should open, right?08:56
tsdgeosdo you see any reason it should not?08:56
* Cimi reads code08:56
tsdgeosneed to reboot, back in a sec09:01
mzanettiSaviq: hehe... that was too easy after all: https://code.launchpad.net/~mzanetti/unity/dont-halt-on-test-failure/+merge/15760709:04
Saviqmzanetti, orly?09:07
Saviqmzanetti, `make check` doesn't need it09:07
Saviqmzanetti, it only runs make test09:07
mzanettioh right... we changed that... lemme check09:08
mzanettiSaviq: updated09:10
Saviqmzanetti, you should update runtests for the move from in-tree builds to builddir09:15
mzanettiSaviq: where is builddir?09:15
Saviqmzanetti, "builddir"09:15
Saviqmzanetti, if you go ./build09:15
Saviqmzanetti, it will build in builddir now, out of tree09:16
mzanettiSaviq: runtests supports in-source and builddirs in the root09:16
mzanettie.g. source-dir at ..09:16
Saviqmzanetti, ah, so ../runtests.sh, got it09:16
mzanettiSaviq: yeah... jenkins builds in a subdir too... so thats supported already09:17
mzanettiSaviq: the only thing not supported is build dirs totally outside the source... that would require passing a parameter for the src-dir I guess09:17
Saviqmzanetti, we could always just .cmake it09:18
Saviqmzanetti, but it's fine for now09:18
=== mmrazik is now known as mmrazik|afk
Cimitsdgeos, it's weird09:35
Cimitsdgeos, I'm trying the one in the ubuntu-ui-toolkit demos09:35
Cimitsdgeos, not sure how is it supposed to behave09:35
tsdgeosi pinged zsombi in #ubuntu-touch and got half an answer :-/09:35
tsdgeosASSERT failure in QList<T>::at: "index out of range", file /home/tsdgeos/qt5/build/qtbase/include/QtCore/../../../../qtbase/src/corelib/tools/qlist.h, line 45209:36
tsdgeoswops :D09:36
nic-doffayHey guys, still getting build errors with my branch, Jenkin's reports a success though. https://pastebin.canonical.com/88624/09:42
tsdgeosnic-doffay: that's build -s ?09:43
nic-doffayIt is tsdgeos09:43
tsdgeosnic-doffay: is it a clean build?09:44
nic-doffayIt was tsdgeos09:45
tsdgeosnic-doffay: can you try build -s -c just to make sure?09:45
nic-doffaysure, I'll let you know when it's completed.09:45
tsdgeosnic-doffay: you have the version of build_unity that contains stuff like09:47
nic-doffayI'm not sure tsdgeos09:47
tsdgeosnic-doffay: open the file and check the first lines of the file09:48
nic-doffaytsdgeos, that's the version I have.09:48
tsdgeosnic-doffay: you were on quantal?09:49
nic-doffaytsdgeos, correct09:49
tsdgeosmaybe there's a linking issue in the hud code in quantal09:49
nic-doffaytsdgeos, why would it affect quantal but not raring?09:53
tsdgeosit might09:53
tsdgeospulse libraries may have different libs linked in or whatnot09:53
nic-doffayWho should I talk to about this?09:55
nic-doffaytsdgeos, ^09:56
tsdgeosnic-doffay: try pete-woods09:57
tsdgeosor tedg when he wakes up09:57
nic-doffayThanks tsdgeos, will do.09:57
tsdgeosnic-doffay: otoh maybe you want to update to raring?09:57
nic-doffayHow stable is it atm tsdgeos ?09:58
tsdgeosworks fine here09:58
nic-doffayCool, I'll just update then.09:58
Cimion Tile.qml, there's a GridView.onRemove, what's that?10:24
tsdgeosSaviq: about the carousel-listview crash, at the moment i'm doing the evil eyes at dee-qt, can't prove is its fault (yet) but it's doing nasty stuff afaics10:27
Cimitsdgeos, where is it crashing?10:28
tsdgeosCimi: deep inside modelview code in Qt10:28
SaviqCimi, http://qt-project.org/doc/qt-5.0/qtquick/qml-qtquick2-gridview.html#onRemove-signal10:28
CimiSaviq, yes but why inside tile?10:29
CimiSaviq, it is catched if tile has a gridview as direct parent?10:29
SaviqCimi, not a direct one, any parent10:30
SaviqCimi, but indeed it should probably be in http://qt-project.org/doc/qt-5.0/qtquick/qml-qtquick2-gridview.html#remove-prop instead10:30
SaviqCimi, Gerry should be able to comment on why he did put it in Tile10:31
CimiSaviq, also, line 86 of that file, it changes the GridView.delayRemove property of AbstractButton?10:32
SaviqCimi, GridView is an attached property of GridView delegates10:33
SaviqCimi, so yeah, scratch my "any parent" comment10:33
CimiSaviq, what will happen if I use Tile outside a gridview?10:33
SaviqCimi, so changing GridView.delayRemove on root means it will change delayRemove on that10:33
SaviqCimi, http://qt-project.org/doc/qt-5.0/qtquick/qml-qtquick2-gridview.html#delayRemove-prop10:33
Cimijust warnings I suppose10:34
SaviqCimi, it will complain about missing GridView10:34
Cimiis that ok to accept?10:34
SaviqCimi, the Tiles were only ever supposed to be GridView delegates10:34
SaviqCimi, so sounds like yeah, it's fine10:35
=== alan_g is now known as alan_g|tea
Saviqtsdgeos, about https://code.launchpad.net/~michihenning/unity/declspec/+merge/157578/comments/34500910:42
Saviqtsdgeos, we'll be moving/splitting the shell under "src/shell", "qml/shell" folders10:42
Saviqtsdgeos, and the rename I believe is to prevent naming confusion10:43
tsdgeosmakes sense10:43
Saviqtsdgeos, so yeah, all source goes under src/10:43
Saviqall (some? we'll have to see) QML+JS goes under qml/10:43
Saviqpublic headers go under include/10:44
=== alan_g|tea is now known as alan_g
SaviqMacSlow, ping10:48
mzanettiIl'l modify the jenkins ci job now.. please ping me if you encounter unexpected failures10:49
Saviqmzanetti, yay10:51
MacSlowSaviq, pong10:54
=== MacSlow is now known as MacSlow|lunch
tsdgeosqt 5.1 alpha out11:15
Cimitsdgeos, did we change something to the build system?11:30
Cimitsdgeos, after doing build in root of unity phablet11:31
Cimiand I go to qmluitests, run make test or make testTile (working on Tile), it complains of no rule or target11:31
tsdgeosCimi: yes, we changed some things11:31
tsdgeosCimi: we have a builddir now11:31
tsdgeoscd builddir11:31
tsdgeosand make test there11:31
Cimimm where?11:32
Cimiin builddir root?11:32
Cimitsdgeos, how to run qmluitests?11:32
tsdgeossame way11:32
tsdgeosmake qmluitests11:32
tsdgeosin the builddir11:32
tsdgeosmzanetti: wrong link https://code.launchpad.net/~unity-team/unity/phablet.test_responsiveflowview/+merge/157343/comments/34511211:41
tsdgeosmzanetti: still contains s-jenkins on it11:42
Cimiquestion. I am in Tile.qml and I need to test the animation that happens when I remove a tile11:42
Cimiwhat I do is getting the item before removing the element (var ItemRemoved = gridView.itemAt(units.gu(4), units.gu(4))11:42
Cimithen I do model.remove(0,1) (I can see the element removing with the animation)11:43
Cimibut the line after that is tryCompare(itemRemoved, "opacity", 0) which is undefined because the item was remvoed11:43
Cimiwhat's the best way to deal with that?11:44
mzanettitsdgeos: oops. my IRC client was detached... anyways, its fixed now11:54
mzanettianyone a small test review? https://code.launchpad.net/~mzanetti/unity/phablet-tile-tests/+merge/15763011:55
Saviqdandrader, you around?12:02
Cimimzanetti, I was testing Tile as well, do you know how to test the gredview remove?12:08
mzanettiCimi: in a meeting right now...12:08
mzanettiI'll ping you when done12:08
sil2100didrocks: hi! I still see quite a lot of failures on the generic autopilot job, I'll be looking into those now12:18
=== MacSlow|lunch is now known as MacSlow
smspillazsil2100: are we parallelizing the compiz ci builds ?12:27
sil2100smspillaz: now I have no idea ;/ Maybe fginther` would know12:31
sil2100fginther`: ping, you around?12:31
=== alan_g is now known as alan_g|lunch
WaltherNote to developers: Unity doesn't seem to handle a portrait display too well - when changing desktops the "splash screen" that shows consists of landscaped screens, in Appearances / background screen the display is in landscape mode, etc12:47
smspillazWalther: file a bug against compiz stating that the wall plugin doesn't handle the portrait case12:51
smspillaz(for the live previews)12:51
=== _salem is now known as salem_
Walthersmspillaz: thanks for the specifics12:52
didrockssil2100: excellent! thanks. I think it's the 100scopes jobs you are talking about12:55
mzanettiCimi: hey13:03
smspillazsil2100: hey, so just do double check13:07
smspillazsil2100: if you do something like ctest -D ExperimentalMemCheck -R .Xorg. all those tests that are being marked as failed in valgrind in CI pass locally right ?13:08
* cyphermox publishes indicators to raring13:08
didrockscyphermox: \o/13:08
sil2100smspillaz: let me double check that - but I was doing make test before with integration tests enabled and all was green13:10
smspillazThere is a condition I know about which might cause them to fail under valgrind13:10
smspillazbut I haven't seen it recently13:10
mzanetticyphermox: hey... any news regarding the autopilot-qt tests? we would need to add some more tests but would like to get that one merged first13:11
cyphermoxmzanetti: remind me what you mean by that?13:12
mzanetticyphermox: https://code.launchpad.net/~mzanetti/autopilot-qt/add-tests/+merge/15369513:12
cyphermoxmzanetti: ok, just waiting for ubuntu-release to review13:14
cyphermox(approving the bug)13:14
seb128cyphermox, thanks for publishing indicators!13:15
mzanettiSaviq: https://jenkins.qa.ubuntu.com/job/unity-phablet-ci/13:18
mzanettiisn't it beautiful?13:18
Saviqmzanetti, I love our test count :D13:18
Saviqkgunn, ^13:18
mzanettionly 250 to go until we overtake Mir :D13:19
* smspillaz takes his 1414 tests and goes home13:19
kgunnmzanetti: Saviq ...that feels pretty good to see :)13:19
sil2100smspillaz: I got one failure when running the tests with the command you specified, let me paste it to you on priv13:19
davmor2Saviq: ah you have 666 then :)13:24
Saviqdavmor2, no, not really ;)13:24
davmor2Saviq: that or 133713:24
Saviqdavmor2, we just went up from ~20 to ~160 in one go :D13:24
fginther`sil2100, smspillaz, compiz-ci should be built in parallel13:29
sil2100fginther`: thanks!13:31
davmor2Saviq: Ha nice :)13:35
=== alan_g|lunch is now known as alan_g
mzanettiSaviq: I just came across all those small files like VideosFilterGrid, VideosCarousel13:49
mzanettiSaviq: they basically don't add functionality in respect to FilterGrid, Carousel etc13:49
Saviqmzanetti, yeah, they just use it13:49
sil2100didrocks: so anyway, currently the generic autopilot job is for 100scopes, yes? Is it high-priority to get those fixed, or is it pushed back now that 100scopes didn't go in?13:49
mzanettiSaviq: however, if we're clever we _could_ run the FilterGrid tests ON them and see if it still works13:49
Saviqmzanetti, true13:49
didrockssil2100: no, the generic autopilot job is… generic13:50
didrockssil2100: so used for oif, indicators, unity, both raring and head, as well as 100scopes13:50
sil2100didrocks: ah, so it varies between builds13:50
didrockssil2100:right ;)13:50
sil2100Good to know *notes it down*13:50
didrockssil2100: you can look at the parameters13:50
didrockssil2100: you have the job, release, ppa parameters specified13:50
sil2100didrocks: since I saw that 100scopes was the latest, though the failures are more generic13:50
sil2100i.e. the latest build was from 100scopes13:51
didrockssil2100: yep, maybe we should remerge first unity against it13:51
didrocksdednick: do you have some time for that? unity trunk latest rev has good results… ^13:51
dednickdidrocks: sure.13:53
didrocksthanks :)13:53
didrockssil2100: so, let's look tomorrow's result with those latest tests13:53
=== dandrader is now known as dandrader|afk
dednickdidrocks: r3283 ?13:55
Cimimzanetti,  do you know how to test the gridview remove for the tile?13:55
Cimimzanetti, there is an animation13:56
didrocksdednick: exactly :)13:56
mzanettiCimi: create a GridView and call model.add() and remove() on its model13:57
tsdgeosSaviq: do you have a second to confirm you can reproduce the crash on my simple test?13:57
sil2100didrocks, dednick: some failures will go away since lp:unity had the dash-overlay-button fixes reverted13:57
Saviqtsdgeos, hit me13:57
sil2100Which was causing some failures (the dash maximized bug)13:57
didrockssil2100: yeah, that's my guess, so let's wait for tomorrow :)13:57
mzanettiCimi: and use a tryCompare for the SequentialAnimation in there for finish13:57
Cimimzanetti, yes I did that13:57
Cimimzanetti, but it removes the item13:57
didrockssil2100: but latest time we synced that with trunk, it was a release, not a random #13:57
Cimimzanetti, so I don't have the item anymore13:58
Cimimzanetti, to test the funcion13:58
sil2100Ok ;)13:58
sil2100Let's see where it will lead us then13:58
tsdgeosSaviq: https://chinstrap.canonical.com/~aacid/qml_model_crasher.tar.gz13:58
mzanettiCimi: right...13:58
Cimimzanetti, http://paste.ubuntu.com/5689471/13:58
seb128sil2100, hey, do you know why https://code.launchpad.net/~seb128/autopilot/correct-majhongg-name/+merge/157372 didn't get merged, didrocks approved it on friday13:58
Cimimzanetti, line 6 fails13:58
mzanettiCimi: you could make sure that the item does not get deleted immediately, but only after the time the animation takes13:58
Cimimzanetti, what?13:59
didrocksseb128: I think fginther` didnt' redeploy head for qa with it?13:59
Cimimzanetti, I believe fakeModel.remove takes the time for the animation also13:59
Saviqtsdgeos, yup13:59
seb128didrocks, oh, ok13:59
mzanettiCimi: remove() blocks? no...13:59
* Cimi puts a console.log14:00
mzanettiCimi: anyways... not really sure how much sense it makes to test this...14:00
Cimimzanetti, you're right14:00
Cimimzanetti, ok14:00
CimiI'll just approve yours then14:00
tsdgeosSaviq: so that's it, i can't find anything you could blame in that code, so it's a Qt bug, not that it matters much, still crashing :D14:00
mzanettiCimi: sorry again for the collision... :/14:00
Cimimzanetti, no worries, I was losing time ttrying to test this :)14:01
Cimimzanetti, you think we should add // for testing14:01
Cimimzanetti, next to objectName = "" in the Tile.qml file?14:02
mzanettiCimi: hmm... I think objectNames are in 99% of the cases used for testing14:02
tsdgeosSaviq: the trick is the double remove and using the listView.count as part of the delegate14:02
mzanettiCimi: besides, if you remove it Jenkins will tell you that it was needed for testing14:02
sil2100seb128: I know about it, I was poking fginther` about it just a few minutes ago14:03
tsdgeosSaviq: so what's the next step, want me to try finding a fix or just report the bug and walk away?14:03
sil2100seb128: and he's on it right now14:03
seb128sil2100, ok14:03
seb128sil2100, shame that you dupped work there :-(14:03
=== fginther` is now known as fginther
sil2100seb128: that's just like a one liner ;p Still have that branch there, need to remove it14:03
Saviqtsdgeos, report, try and fix, please14:04
Saviqtsdgeos, you decide when you've had enough14:04
mzanettiCimi: you had tests for DashBar, right?14:04
Cimimzanetti, yes14:04
sil2100seb128: so no problem ;)14:04
Cimimzanetti, but we're waiting SDK14:05
mzanettiCimi: ah ok... was just wondering why stats still say no tests for it14:05
* fginther growns14:05
mzanettiCimi: do you have a link for the MP?14:05
mzanettifginther: hey ho14:05
* fginther and groan as well14:06
Cimimzanetti, https://code.launchpad.net/~unity-team/unity/phablet.dashBar_bottomswipe/+merge/15084714:06
fgintherdidrocks, the "to_transition" bit is throwing off the auto-merger tools14:08
* fginther starts work on a fix14:09
didrocksfginther: coordinate with cyphermox for the QA stack14:09
didrocksI think it can be transitionned, but better to check14:09
fgintherdidrocks, ack14:09
didrocksat least for autopilot :)14:09
cyphermoxyeah, just need to finish it up14:09
fgintherdidrocks, we really need to treat the 'to_transition' projects as regular projects, we want to keep auto-merge these even if they aren't being built for the daily release14:10
=== dandrader|afk is now known as dandrader
didrocksfginther: no, because look at unity in raring for instance14:11
didrocksfginther: it's still treating lp:unity, not any branches14:11
didrocksfginther: and in the transition guideline, we only move stuff back when the 2 branches diverges14:11
didrocksso at the same time, change lp:unity/raring to raring for instance14:11
didrocksand reenable lp:unity in head14:11
tsdgeosSaviq: https://bugreports.qt-project.org/browse/QTBUG-3055514:11
Saviqtsdgeos, cheers14:12
fgintherdidrocks, hmm. I see your point14:13
didrocksfginther: so, normally, all steps are done in sync ;) I guess it's a little funky until we really get to have this habit14:13
didrocksfginther: btw, I don't remember if I sent it to you: https://wiki.ubuntu.com/DailyRelease/MovingNewRelease14:13
fgintherdidrocks, thx14:14
didrocksyw ;)14:15
dednickdidrocks: https://code.launchpad.net/~nick-dedekind/unity/smart-scopes.merge-trunk-3283/+merge/15766514:20
fginthersil2100, seb128, we figured out why lp:autopilot is not auto-merging, just need to sync some transition items14:22
cyphermoxfginther: do you have sufficient access to attach branches to series?14:22
Cimidednick, are you writing tests for searchindicator of panel, right?14:22
fginthercyphermox, I can try14:23
cyphermoxI'd finish fixing up the autopilot bits that aren't properly split to feature branches?14:23
cyphermoxbasically, there is -gtk, -qt, dbus-test-runner, gtester2xunit, pyruntest, xpathselect and window-mocker14:24
dednickCimi: i'm trying to only write tests at a file level. so Indicators.qml in one test, Panel.qml in another. I havent got on to the panel yet, but i dont think it'll include the searchIndicator at this time. That would be another test case.14:24
fginthercyphermox, and they all need /raring branches? (or some equivalent)?14:24
cyphermoxyeah those need raring branches. basically trunk.13.04 as per tradition14:25
cyphermox(I'm finishing up checking to make sure they don't already have them)14:25
sil2100fginther: thanks!14:25
Cimidednick, I'll write it then14:26
didrockscyphermox: better to check with QA upstream, because it seems there are using another way or versionning though14:26
dednickCimi: be my guest :)14:26
didrocksas per what I'm seeing in autopilot14:26
cyphermoxdidrocks: I'd let fginther or mmrasik deal with the branching and series things though14:27
cyphermoxI certainly don't have access for it14:27
didrockscyphermox: just sync with them for deploying head once ready ;)14:27
cyphermoxdbus-test-runner is the only one that is already split up, just needs that I update the config14:28
cyphermoxdidrocks: what do you mean?14:28
didrockscyphermox: updating the config and deploying both raring/ and head stacks because they start pushing new features in their trunk :)14:28
didrockscyphermox: hence, the "it needs coordination"14:29
cyphermoxyeah yeah :)14:29
didrocksthanks cyphermox :)14:29
cyphermoxfginther: are you going to create the branches, or should I speak with mmrasik before?14:30
fginthercyphermox, I'll work with thomi to get this sorted out, then notify you14:30
fginthercyphermox, didrocks, does a project need to be split (into a maintanence and trunk branch) if there are no new features being developed?14:32
cyphermoxfginther: not really14:32
cyphermoxbut then perhaps make sure we're aware so we remove the project from head assuming it's not going to be in the next release14:32
didrocksfginther: it's upon upstream request, when they want to do develop new features and so, need to split the branches14:32
Saviqtsdgeos, reading more into your test code, it only happens with a QAbstractListModel?14:34
fginthercyphermox, didrocks, and if upstream does not split, we run the risk of accidentally pushing a feature change to the maintenance release, correct14:34
=== alan_g is now known as alan_g|tea
didrocksfginther: right, but they got an email telling to not do that :)14:35
didrocksfginther: so if everyone is rigorous, that shouldn't happen14:35
didrocks(as we have a submitter and a reviewer)14:35
cyphermoxfginther: we do, although that's why I also insist on reviewing all changes for raring :)14:35
fgintherdidrocks, right!14:35
didrocksalso, cyphermox and other people from ~ubuntu-unity are watchdogs ;)14:35
cyphermoxfginther: mzanetti can confirm we've blocked merges in the past :)14:36
Saviqtsdgeos, shouldn't using a ListModel { } expose the same crash?14:36
cyphermoxdidrocks: on that subject, do you have the power to review https://code.launchpad.net/~mzanetti/autopilot-qt/add-tests/+merge/153695  and the FFE for it?14:36
tsdgeosSaviq: not sure, maybe, haven't tried14:36
Saviqtsdgeos, would be good to get rid of the cpp code14:36
didrockscyphermox: I'm not on the release team, but I would say that's not a FFe to add tests ;)14:37
didrocksah, new package14:37
cyphermoxdidrocks: I disagree. it's a feature though one we should be happy to easily include :)14:37
didrockscyphermox: maybe get someone from the release team for a quick ack?14:37
mzanettiSaviq: do you think I'm good enough to judge on the greeter stuff or do you want to join the hangout?14:37
Saviqmzanetti, what about it?14:38
didrockscyphermox: so, once you add that, you will have the integration tests running for the QA stack, right?14:38
cyphermoxdidrocks: I think that's the plan yeah14:38
mzanettiSaviq: well, we have a meeting now on how to fully integrate with lightdm etc14:38
Saviqmzanetti, you're good14:38
tsdgeosSaviq: let me try to kill the cpp code14:38
didrockscyphermox: not sure what I should review, I think you should review the packaging and let in your capable hands. Tell me if you need help to setup the integration tests then ;)14:38
tsdgeosSaviq: but i doubt anyone can shift the blame to my 3 lines of cpp code14:39
Saviqtsdgeos, ;)14:39
Saviqmzanetti, just ping me if you need me14:39
cyphermoxdidrocks: the packaging looked fine although I'm unsure about the same for the package14:39
mzanettiSaviq: ack14:39
didrockscyphermox: sorry, I don't get you, if the packaging looked fine, what about the package?14:40
cyphermoxsame == name :)14:40
cyphermoxthe naming for the new binary :)14:40
cyphermoxdid we ever agree on a naming scheme for the autopilot tests packages?14:41
tsdgeosSaviq: yep, the cpp code is not needed14:41
didrockscyphermox: <binary-component-pkg-name>-autopilot14:41
tsdgeosi'll update the attachment14:41
Saviqtsdgeos, cool14:42
cyphermoxdidrocks: then we're good14:42
cyphermoxI just wasn't sure we had reached an agreement14:42
didrockscyphermox: awesome, maybe ping Laney for a quick release team review?14:42
cyphermoxI just mentioned it in #ubuntu-release14:43
didrockscyphermox: well, I think it's the easiest naming we can fine :)14:43
cyphermoxI'll wait until release team decides and take the time to get some stuff planned for protocol stacks14:44
tsdgeosSaviq: http://paste.kde.org/~tsdgeos/718802/ the new one14:44
Saviqtsdgeos, yup14:44
Saviqtsdgeos, good work14:45
tsdgeostell me that when i fix it :D14:45
Saviqtsdgeos, I will!14:45
didrocksbregma: FYI, fixing bug #1164915 made bug #1108956 to be reverted15:11
ubot5bug 1164915 in Unity "dash doesn't close when pressing <super> on a specific lens" [Undecided,Fix committed] https://launchpad.net/bugs/116491515:11
ubot5bug 1108956 in unity (Ubuntu) "The dash closes when trying to switch from Command lens to home lens" [Low,In progress] https://launchpad.net/bugs/110895615:11
bregmahardly surprising, given the intricacies of all the keypress handling15:11
didrocksyep ;)15:12
Cimimzanetti, on the name property of Test elements, sometimes we use the suffix Test and sometimes not, shall we write a mail on the ML about always adding it?15:13
Cimiof TestCase elements15:14
mzanettiCimi: yeah, feel free to do so. I haven15:14
mzanettiI haven't figured what exactly it changes... but I think it makes it easier to identify failures etc15:14
didrocksTrevinho: when you are going to get nux merged who broke the ABI, please ensure you follow https://wiki.ubuntu.com/DailyRelease/FAQ#My_package_B_depends_on_a_new_symbol_I_just_added_on_A15:15
didrockshey mterry!15:15
Trevinhodidrocks: ah15:15
didrocksmterry: do you think we can get a new Unity publication today? ;)15:15
didrocksTrevinho: basically, it's whenever you do an ABI break15:16
mterrydidrocks, ah let me see15:16
Trevinhodidrocks: fine, thanks15:16
didrocksTrevinho: in fact, it's rather in https://wiki.ubuntu.com/DailyRelease/FAQ#I_need_to_break_an_API.2BAC8-ABI :)15:16
didrocksMirv: still around?15:18
Trevinhodidrocks: so to make nux to trigger a rebuild of unity... what should be added?15:20
didrocksTrevinho: did you read about bumping the build-dep in unity against latest nux?15:20
didrocksTrevinho: and bump the version in debian/changelog of nux?15:20
Trevinhodidrocks: mh, yeah.. I read... ok15:21
mterrydidrocks, published15:21
didrocksmterry: \o/15:21
Trevinhodidrocks: I should add it with debchange -i, irght?15:21
TrevinhoI mean, as a new unreleased version15:22
didrocksTrevinho: no15:22
didrocksAlso, please do think to bump the version in debian/changelog as well (this will be for next release).15:22
didrocksIf there is already an UNRELEASED content, change it from15:22
didrocks'''0.42.1daily83.13.09-0ubuntu1''' (or '''0.42.1-0ubuntu1''')15:22
didrocks'''0.42.2-0ubuntu1''' for instance (you don't need to strip the daily part, if you do, the daily release will readd it at next successful release)15:22
didrocksfrom the FAQ15:22
Cimimzanetti, sometimes we put TestCase/UnityTestCase before the components, sometimes after... guideline?15:22
mzanettiCimi: I don't have a strong opinion on that...15:23
Cimimzanetti, me neither, but let's decide or anarchy will reign :)15:23
mzanettiCimi: well, usually one would put TestCase as the root item15:23
mzanettiCimi: sometimes we need another Item to wrap it..15:23
mzanettiCimi: so for me the logical order would be TestCase first15:24
Cimimzanetti, ok15:24
Trevinhodidrocks: mh, yeah, but there's not an ureleased version in nux, that's why I asked15:24
didrocksTrevinho: I'm adding:15:25
didrocksOr run '''$ dch -i''' and change the version to match this (ensure you have UNRELEASED content and not "raring" or any version in the first line).15:25
didrocksdoes it make sense?15:25
kgunnnic-doffay: hey...wrt defining what app info you need for infographics15:25
Trevinhodidrocks: yes15:25
kgunnnic-doffay: i would recommend following a model like notifications used15:25
kgunnnic-doffay: ^ basically use a test to define/negotiate your api15:26
Trevinhodidrocks: also, if there are multiple abi changes, I don't want to change version to 4.0.1 -> 4.0.2 ... WHat about -> instead?15:26
didrocksTrevinho: why not 4.0.1?15:26
didrocksTrevinho: you know, normally breaking ABI would even force bumping the soname :)15:27
Trevinhodidrocks: If I've two branches changing the ABI in the same minor release15:27
didrocksTrevinho: if you land them in the same day, bumping once is enough15:27
Trevinhodidrocks: ok15:27
didrocksTrevinho: as just one branch for unity is enough15:27
didrocksTrevinho: basically, everything needs to land in the same day, before next daily15:27
didrocksTrevinho: FYI: https://wiki.ubuntu.com/DailyRelease/FAQ?action=diff&rev2=17&rev1=1615:28
Trevinhodidrocks: yeah, that was my guess... But I was thinking to edge cases15:28
didrocksI would prefer we avoid version of 4kms long :)15:28
Trevinhodidrocks: nice15:28
Saviqkgunn, nic-doffay, we need a chat with pete-woods before then to agree on an overall architecture15:30
kgunnyep...just asked Pete to join15:30
kgunnthostr_: pete-woods just to get on the same page15:31
kgunnI was proposing nic-doffay use Saviq's model for capturing frontend/backend interface15:31
kgunnlike so https://code.launchpad.net/~saviq/unity/phablet.notification-interface-tests15:32
=== alan_g|tea is now known as alan_g
kgunnpete-woods: nic-doffay of course not done in a vacuum...whatever you 2 can negotiate up first, as Saviq said, agree on some basic overall arch15:33
thostr_kgunn: yes, we can use that, but after iteration 0 as discussed on Friday15:33
kgunnthostr_: sure15:33
kgunniteration0 = quick and dirty15:33
kgunnintegration to learn some things...only to turn around and capture in the test15:34
thostr_kgunn: yes. but if guys have a good unterstanding I wouldn't be opposed to start with iteration 115:34
thostr_kgunn: I think we are in learning phase... I'm open let's see which approach works better in the end (having iteration 0 vs not)15:34
thostr_kgunn: as we learned: most important part there is anyway people actually talking to each other15:35
Saviqthostr_, kgunn it also depends on the scope (pun intended) of the api15:35
kgunnthostr_: i think nic-doffay did at least have a set of data he needs from apps...15:35
thostr_Saviq: true15:36
Saviqthostr_, kgunn, infographics should be very thin, it's a list of values, a label and a button after all15:36
nic-doffaykgunn, yeah I've just seen a list of wanted data.15:36
nic-doffayWho shall I share the doc to?15:36
kgunnnic-doffay: pete-woods15:36
pete-woodsnic-doffay: I'm obviously interested15:36
Saviqnic-doffay, /me, too15:37
nic-doffayOk great, I'll share that over now.15:37
pete-woodsbasically what I'm expecting to create is a service with two APIs15:39
pete-woodsone for the infrographic to interrogate15:39
kgunnnic-doffay: can you add what i presume is a link to this doc ?15:39
pete-woodsand another for apps to register what they have to say15:39
kgunnnic-doffay: ^15:40
nic-doffaySure kgunn15:40
Cimimzanetti, /Panel/IndicatorItem.qml [EASY]15:44
Cimitest position of label / icon15:44
Cimimzanetti, reading the coordinates?15:44
nic-doffaykgunn, it's there already15:45
nic-doffayright at the bottom.15:45
nic-doffayUnder the greeter tab.15:45
nic-doffay(In that doc you linked)15:45
kgunnnic-doffay: yep...saw it...just didn't know if it was a dup15:45
nic-doffaypete-woods, Saviq https://docs.google.com/a/canonical.com/document/d/1VajNkWbBH61iVixXJAmOvNGiG__GWQTMXGNOZijXWJw/edit#heading=h.dxyj97l61sl715:46
Saviqpete-woods, I love that high-level architecture! :D15:46
nic-doffaypete-woods, page 20 is what data the designers want to access for the infographics.15:46
nic-doffayWhich will influence the display.15:46
Saviqnic-doffay, I think one interesting question is: who decides on the colour of the infographic?15:47
pete-woodsyes, I was thinking about this too, do apps get to say what color their notifications are?15:47
Saviqpete-woods, s/notifications/infographics/15:48
Saviqpete-woods, to avoid confusion15:48
sil2100seb128: fginther merged in that mahjongg-rename branch15:48
seb128sil2100, fginther: thanks15:49
nic-doffayI guess we should start by adding the infographics where ever they need to be in the code, then take things from there.15:51
nic-doffaySaviq, could you direct me with this?15:51
Saviqnic-doffay, on the contrary, you should start creating the component with some mock data behind it as a component15:52
pete-woodsdefinitely this was round15:52
Saviqnic-doffay, and build tests for it15:53
Cimiif someone approves this, I don't have to make newer branch depending :) https://code.launchpad.net/~unity-team/unity/phablet.test_Panel-searchIndicator/+merge/15767915:53
pete-woodsnic-doffay: I would hope to work with you to get an API that gives you what you need right from the start15:53
nic-doffaySaviq, ok great.15:53
pete-woodseven if it gives nonsense data15:53
CimiSaviq, I put the first test for SearchIndicator under a Panel subdirectory15:53
SaviqCimi, k15:53
nic-doffayAgreed pete-woods the sooner that we have that, the better even if it's nonsense.15:53
CimiSaviq, if you want, I can move all components tests under a Components dir15:54
SaviqCimi, separate MP, but yeah, you can prepare one that moves the tests in the correct paths15:54
CimiSaviq, obviously a separate one15:54
Cimidoing now15:55
Saviqpete-woods, can you set up a quick (half hour should be enough) hangout tomorrow morning?15:55
Saviqpete-woods, so that we look at it from a birds-eye view and think what's needed?15:55
pete-woodsSaviq: who do you want in the hangout? you me and nic-doffay?15:55
pete-woodsor more?15:55
Saviqpete-woods, yup, should be enough15:55
Saviqpete-woods, I'll then sit with nic-doffay and prepare the tests for the API15:56
mzanettiCimi: the question still valid?15:56
Cimimzanetti, yes15:56
Cimimzanetti, the question was what you meant there15:57
Cimimzanetti, on that file15:57
Cimimzanetti, my MR is for SearchIndicator15:57
nic-doffaySaviq, about the PageHeader control test. Shall I keep that assigned to me and just prioritise the Infographics work?15:57
mzanettiCimi: hmm... Saviq wrote that... I think he meant to check if the label is only visible if non-empty etc15:58
Cimiok will think about it15:58
dandradergreyback, ping15:59
greybackdandrader: pong15:59
dandradergreyback, about the "close apps from dash" story: seems you made it so that if the user taps anywhere over the app thumbnail/tile (as opposed to only over its close icon) it will cause the application to close16:00
dandradergreyback, what the reasoning behind it?16:00
pete-woodsSaviq: where do you see tests for the "infrograpic" API existing?16:01
mzanettitsdgeos: would you consider all files in HUD tested sufficiently?16:01
greybackdandrader: was design request16:01
dandradergreyback, ahh... so we did have some design on that feature after all :)16:01
dandradergreyback, And how do I dismiss the close mode?16:02
dandradergreyback, switch do another dash?16:02
greybackdandrader: by request I mean I was showing it off to designers and was told to do that :) Nothing written down unfortunately16:02
greybackdandrader: I recall that in delete mode, the other visible areas of the dash are dimmed. Then you could dismiss delete mode by tapping darkened bits (including the header) or opening launcher/indicators, or switching to different lens16:03
greybackdandrader: that's from memory. Mika was the UX designer I was working with, and Jouni the visual designer.16:05
pete-woodsSaviq: because obviously if I'm writing an API someone else has already written tests for that's like developer heaven ;)16:05
greybackmzanetti: too late :)16:06
greybackmzanetti: tsdgeos has a physical power-off timer switch on his machine that fires at 6pm :)16:06
mzanettigreyback: what?16:06
mzanettigreyback: I bet I can reach him using mzanetti@kde.org :D16:07
greybackmzanetti: but that'll not be the tsdgeos we know and.. tolerate. He'll be some KDE mutant twin16:07
mzanettigreyback: that fits me quite well :P16:08
greybackmzanetti: actually while I got you, in testing ListViewWithPageHeader, I was having some trouble moving the Flickable a tiny bit. I know of your mouseFlick method, but I couldn't make it move the content just a little bit. Any tip?16:09
greybackmzanetti: I can emulate a small flick manually with a bunch of mouseMove calls, but it ain't pretty16:10
mzanettigreyback: I fear thats the only way :/16:10
greybackmzanetti: ok, it works so I'll leave it16:11
mzanettigreyback: iirc there are some comments in Flickable's codebase that describe this... you have to interpolate your flick yourself, otherwise it won't do16:11
greybackmzanetti: yep, I've been reading hte source too, saw that16:11
greybackmzanetti: ok, that's 3 hours too much time I've spent at that :(16:11
mzanettigreyback: probably it would make sense fix our mouseFlick() method to enable it for that16:11
greybackmzanetti: yep, I'm considering it16:12
greybackmzanetti: would be a fun bit of maths :)16:12
mzanettiyes, Herr Doktor16:13
dandradergreyback, another thing: why did you also add a CloseIcon to Tile (besides RunningApplicationTile)16:13
greybackmzanetti: :P16:14
greybackdandrader: in case in future it was wanted. Say to uninstall applications.16:14
greybackdandrader: no need for you to agree with that tho16:14
kgunnpete-woods: i don't think there is an API....and that was the idea, nic-doffay could help you out in a way...while at the same time defining the api :)16:15
=== dandrader is now known as dandrader|lunch
kgunnpete-woods: oops...by API...i meant both API & free test :)16:15
greybackdandrader: note there's 1 bad thing I did in that code. I placed the close button with negative anchor margins. That can sometimes cause redraw issues when the delegate is being animated, as I discovered later.16:16
pete-woodskgunn: I'm not sure I follow? I guess I was expecting to be implementing an API that nic-doffay would be interrogating for infrographic data16:16
* mzanetti is sad that now that we have fancy statistics noone proposes any MP's any more :D16:16
Saviqnic-doffay, we'll take over the PageHeader test16:17
mzanettiSaviq: should I finish that one?16:17
pete-woodskgunn: and that I'd have to define another API for app developers to register their information to appear in the infrographic16:17
Saviqmzanetti, sure, have at it16:17
nic-doffayOk Saviq, shall I delete the MP and make a branch on unity-team?16:17
nic-doffaypete-woods, that was the impression I got too :)16:17
Saviqnic-doffay, mzanetti'll take care of it16:17
nic-doffaySaviq, ok.16:18
kgunnpete-woods: i'm overloading terms...i just meant the interface between infographics & backend16:18
mzanettinic-doffay: no problem. I'll just pull yours and continue on that16:18
nic-doffaycool cheers mzanetti16:18
Saviqmzanetti, you can always retrigger all of the outstanding ones :D16:22
mzanettiSaviq: I think they need to be merged with trunk again to pass ci16:23
Saviqmzanetti, right16:23
Saviqpete-woods, the shell-facing API tests will be written in QML and live in lp:unity/phablet (where your API for infographics will later live, too)16:25
Saviqpete-woods, they will be very small, though16:26
=== salem_ is now known as _salem
Saviqpete-woods, just verifying that the API exposes all the needed properties / methods / whatever16:27
pete-woodsSaviq: would you want the service implementation in there too? I'm expecting it to be a dbus service16:27
Saviqpete-woods, I don't care it being a dbus service or not ;)16:27
sil2100ChrisTownsend: hi! About the SRU timeline for quantal and precise16:27
Saviqpete-woods, and the implementation itself should live outside of lp:unity/phablet16:28
sil2100ChrisTownsend: we will try preparing releases till the EOW, but from what Didier said, it might take a while for them to get uploaded16:28
sil2100Up to 1 month16:28
pete-woodsSaviq: okay, so something like a Qt API living somewhere else? and a QML API in the unity tree?16:29
Saviqpete-woods, no, the APIs will all live in the unity tree16:29
Saviqpete-woods, both shell-facing and app-facing16:29
Saviqpete-woods, but the implementation of the service16:29
Saviqpete-woods, will, if possible, live outside16:30
tedgSaviq, The implementation of the service needs to live in the same repo as the APIs to it are used.16:30
pete-woodsSaviq: so it will become very important we sync the version of the service to the version of unity then16:30
tedgSaviq, We don't want things like dbus interfaces spanning repos.16:30
Saviqtedg, of course we don't, but we can abstract from it16:31
tedgSaviq, So you want lib -> lib, lib -> service -> lib, lib -> lib ?16:31
Saviqtedg, but the unity repo might only include mock implementations of the service16:32
tedgSaviq, Basically so the dbus interface is with teh service.16:32
pete-woodsSaviq: if you want it in the unity tree just to make sure it has test coverage, you need not worry, wherever I put it, it will be well tested16:32
Saviqpete-woods, no it's not that, it's about syncing changes in the API  - we want changes to API to break the shell straight away16:33
Saviqpete-woods, not only when it's built in distro / PPA16:33
pete-woodsSaviq: is the shell even going to be using the API?16:34
Saviqpete-woods, some API, yes16:34
Saviqpete-woods, for the shell it's going to be a QML API, but built on top of some C++ one16:34
tedgSaviq, But, let's say there's another API that's delivered by the service's repo.  When that changes it'd break your build the same.16:35
pete-woodsSaviq: of course that's how I'll build it, but I'm just interested which part of the API you see the shell using?16:35
tedgSaviq, They only way you get that is to copy the entire world into the Unity repo.16:35
Saviqtedg, not if the Unity repo defines just the interfaces that the service then implements16:36
tedgSaviq, So you want to supply the -dev package and have the service provide the lib package?  That seems a bit insane to me.16:37
Saviqpete-woods, not sure what parts do you have in mind16:38
Saviqtedg, I'm not opposed to implementation living within the Unity repo, but we don't want, as you put it, to "copy the whole world" there16:39
pete-woodsSaviq: well I see it very superficially as a service with an IN and an OUT16:39
Saviqpete-woods, yeah, the OUT part is what the shell will be interested in16:40
pete-woodsSaviq: the OUT is very much the territory of the infrographic16:40
Saviqpete-woods, and the IN part is what the apps will be interested in16:40
pete-woodsSaviq: which isn't part of the shell?16:40
pete-woodsthe OUT I mean16:40
tedgSaviq, At some point there is going to be an interface between your repo and the other repo.  That can/will break.  Just making sure you realize that :-)  We can put as many fancy words around as we want :-)16:41
pete-woodsthe infrographic is in the greeter16:41
Saviqpete-woods, it's actually in the welcome screen16:42
pete-woodsSaviq: surely that shouldn't be part of the shell?16:42
Saviqpete-woods, depends16:42
pete-woodsSaviq: what about when we change user?16:42
Saviqpete-woods, that's actually the login screen16:42
Saviqpete-woods, that design folks is thinking of splitting out of the welcome screen16:42
Saviqpete-woods, but anyway I was simplifying by including the greeter in the shell16:42
kgunnupgrading to raring...16:42
Saviqit's not part of the shell as in the session shell16:43
Saviqbut part of the shell as in the system shell16:43
Saviqtedg, sure, but we can try and minimize the impact16:43
pete-woodsSaviq: okay, well if the shell is consuming the OUT API, I can understand why you care about it16:44
pete-woodsSaviq: I would tend to agree with tedg that you can't expect to have all your APIs in the shell tree, though16:44
pete-woodsSaviq: for example if I implement the service out of tree, then it's the DBUS API that breaks, and you only notice at run / test time, instead of compile time16:45
Saviqpete-woods, either way it's going to be a Unity API and whether it's going to be consumed by the shell or the greeter doesn't really matter16:45
Saviqpete-woods, tedg why could the interfaces (as in abstract classes) not be with the shell tree16:46
tedgSaviq, How do you change them in sync with a release of the service?16:46
pete-woodsSaviq: I mean tbh I'll do it whichever way you think works best, I just worry about missing API breakages by keeping it all separate16:46
Saviqtedg, same as usual, just pray ;)16:47
Saviqtedg, and depend on the libunity-api version you built the service against16:47
tedgSaviq, Well, we don't pray.  We version the package and make sure the binary versions are the same through packaging requirements.16:47
tedgI mean, sure, it just seems like it's making life more difficult.16:48
tedgThe goal being "make unity never break" which I think we've shown really isn't achievable anyway.16:49
Saviqtedg, the thing is that, if the interfaces live with the shell, before you can build the service, you'll have to make sure that the updated APIs don't break the shell16:50
Saviqtedg, and you can't actually check in any API change that would, 'cause CI would prevent that16:50
nic-doffaySaviq, would infographics be considered a part of /Components?16:51
Saviqnic-doffay, not really, /Greeter, I'd say16:51
nic-doffaySaviq, doh! didn't see it!16:51
tedgSaviq, All you're guaranteeing is that Unity C++ remains buildable, not that it works or links properly.  It seems like a silly optimization to me.  If it makes you happy, that's fine.16:53
Saviqtedg, don't get me wrong, first of all, I'm not saying "do it like that!", I'm trying to find a solution16:56
Saviqtedg, second, I'm not opposed to putting the service implementations with the shell, but that's not going to be possible at times (like scopes, for example)16:57
pete-woodsSaviq: eventually you have to deal with API breakages outside your package, in my experience you control the versions of your dependencies in CI to either ensure you have the bleeding edge ones, or stable ones16:58
=== _salem is now known as salem_
ChrisTownsendsil2100: Ok, thanks!  Any idea why the delay in the uploads?17:09
sil2100smspillaz: yay! It went through CI correctly this time \o/17:09
sil2100smspillaz: do you mind that I approved the quickfix merge globally?17:09
sil2100ChrisTownsend: didrocks said he will be only able to do the uploads during his next patch pilot, which is in 4 weeks - we might try poking other people, but in the worst case Didier will take care of it in a month17:10
Saviqpete-woods, sure, but our current issue is that if the service defines and holds the API definition, it can be changed at will17:10
Saviqpete-woods, and we can try to alleviate that with versioning, but the fact is that when that happens shell development will be blocked until the API change is addressed17:11
pete-woodsSaviq: only if some joker looks after the API, aren't we all supposed to have, like, well defined ABIs?17:11
pete-woodssurely I should be saying, hey Saviq, API breakage coming..17:12
ChrisTownsendsil2100: Ok, thanks, makes sense.  Really hope it can make it sooner since it seems to have a pretty big impact on Steam users and other apps that use undecorated windows.17:12
Saviqwhereas if the interfaces are defined with the shell, it's the shell / libunity-api that will be changed first17:12
=== alan_g is now known as alan_g|life
pete-woodsSaviq: well you don't really have to convince me, as I'll just go along to be honest17:13
Saviqpete-woods, sure I do ;) I'm fishing, here, too17:13
Saviqpete-woods, we definitely need a discussion (and soon)17:13
pete-woodsSaviq: I'd rather we all pull in the same direction even if I don't agree with it17:14
Saviqto come up with a plan17:14
tedgSaviq, But if there is an API or ABI bump, and the packages are parallel installable, development wouldn't be stopped until someone removes the old package, right?17:15
tedgSaviq, So shouldn't our goal be to make everything parallel installable?17:15
tedgSaviq, And that would, in effect, solve your problem.17:15
tedg(and others)17:15
pete-woodstedg: is it common that packages are parallel installable?17:15
pete-woods(I have never heard of this feature)17:16
CimiI need a little help with a Makefile I guess lp:~unity-team/unity/phablet.moving_tests17:16
tedgpete-woods, Not always, but they can be.17:16
Cimithe Hud test doesn't work with my new hierarchy17:16
tedgpete-woods, It's something we've almost got with hud, but I found a bug in it last week :-(17:16
Cimiwho can help me and have a little look?17:16
Saviqtedg, if you upgrade a package in distro, will the old version still be available for installation?17:17
Saviqtedg, surely you don't mean that the package will provide multiple versions?17:17
tedgSaviq, If someone depends on it, and then when no one depends on it, for about a week.17:17
tedgSaviq, No, the binary package.17:17
tedgsource -> bin1, then in the future, source -> bin2.  bin1 will live as long as there's a dep.17:18
tedgInteresting question on build-dep though....17:18
tedgLet's ask.17:19
mzanettiSaviq: is it wanted that the SearchHistory stores stuff in lowercase or did the test reveal a bug?17:20
Saviqmzanetti, wanted17:20
Saviqmzanetti, well, both17:20
Saviqmzanetti, it should display the exact string entered (stripped)17:20
mzanettiI typed "Humppa"17:20
Saviqmzanetti, but compare lowercase17:20
mzanettiand model.get(0).query returns "humppa"17:20
Saviqmzanetti, bug, then, I'd say17:21
Saviqmzanetti, just tested on the phone17:22
Saviqmzanetti, it stored the same case17:22
Saviqmzanetti, looking at the code, it doesn't even do case-insensitive comparison17:24
mzanettiSaviq: yeah... thats why I asked...17:24
Saviqmzanetti, so to that extent it's broken, but it does store the same case query17:24
mzanettifound the issue17:24
mzanettikeyClick("H") only presses "h", not shift17:25
Saviqtedg, yeah, if we can build against an old version of the API, and everything that came out of the corresponding source package is kept17:25
Saviqtedg, then we should be good either way17:26
tedgSaviq, I put a question in #ubuntu-devel to ask about how that purge happens.  It's possible from a technical perspective, but not sure how fast the archive garbage collector detects garbage.17:26
Saviqtedg, but the other point was that keeping the APIs separate from the implementations would allow for drop-in replacements17:27
tedgSaviq, Hoping build-deps are considered, but I'm not 100% sure.17:27
tedgSaviq, Sure, that's what Provides: are for :-)17:27
Saviqtedg, well, yeah, but then you have to copy the interface between implementations17:28
Saviqor at least depend17:28
Saviqon some "blessed" implementation17:28
tedgWhich is really the same thing as you'd have to depend on the interface in Unity.17:28
tedgIt doesn't matter where the interface is, you'd have to depend on it either way.17:28
tedgPerhaps I'd like a drop in replacement for Unity?  ;-)17:29
=== salem_ is now known as _salem
=== _salem is now known as salem_
Saviqtedg, we did that already! ;)17:32
=== dandrader|lunch is now known as dandrader
=== jhodapp is now known as jhodapp|bbiab
Saviqaaanyway - we definitely need to agree on something, and soon...17:38
mzanettiSaviq: https://code.launchpad.net/~mzanetti/unity/phablet-pageheader-tests/+merge/15771117:47
Saviqmzanetti, go away, it's past your bedtime (and mine, too, trying to get up for over an hour now ;P)17:47
mzanettihehe... yeah... I just finished this task17:47
mzanettiwill go away now17:47
mzanettiSaviq: btw... I discovered a bug in the History Popup. Seems to be the popuputils not our code. couldn't write a testcase for it because the popup is not a child of any of our objects17:48
mzanettibut if you make the window small enough that the popup doesn't fit below the textfield, but next to it, clicking on it doesn't work any more :D17:49
mzanettiwill  report a bug tomorrow17:49
mzanettisee you all tomorrow17:50
Saviqmzanetti, cheers17:50
Saviqmzanetti, and I think it's actually our bug17:51
mzanettiis it17:51
Saviqmzanetti, there's an InverseMouseArea around the popup17:51
mzanettiSaviq: well, clicking the popup does work, and the popup closes, but it doesn't restore the text17:51
Saviqmzanetti, but it only supports the popup being below the text input17:51
Saviqmzanetti, yeah, because you're actually dismissing the popup17:52
Saviqnot clicking it17:52
Saviqmzanetti, anyway, have a good evening17:52
mzanettiyep... unfortunately I couldn't really test it as I have no chance to findChild() the popup17:52
Saviqfreakin' qtcreator17:53
tedgSaviq, So it seems that the garbage collection process is an actual human, and they check the build-deps.  So as long as we can make things parallel installable, you should be good.18:10
Saviqtedg, really? a human? yikes18:10
* tedg is a bit surprised it is a human, but glad it's not him.18:10
tedgSaviq, Yeah, I know.  They're messy and hard to duplicate!18:11
davmor2Hey guys music still isn't opening in RB unless RB is open.  With RB closed. Open Dash on Music Lens select a track.  expect RB to open and track to start playing Raring is the only system not doing this18:45
=== dandrader is now known as dandrader|afk
=== dandrader|afk is now known as dandrader
=== jhodapp|bbiab is now known as jhodapp
=== salem_ is now known as _salem
=== _salem is now known as salem_
tedgSaviq, Okay, more info.20:22
tedgBasically everything will get blocked in proposed.20:22
tedgSo let's say there's a libhudclient2 but unity is still depending on libhudclient120:22
tedgUntil Unity migrates from libhudclient1 to 2, hud won't land nor will unity land.20:23
tedgEverything ends up waiting for a complete change set.20:23
tedgSo if you're developing Unity, until Unity trunk has the merge to update to libhudclient2, libhudclient1 will always be available in the archive.20:24
tedgOnce trunk has it, you should update your dev branch.20:24
tedgBut I'm guessing that is uncontroversial.20:24
Saviqtedg, got it, a followup question - how does library SONAME relate to the API version, then?20:31
tedgSaviq, It doesn't.  It only relates to the ABI version.20:31
tedgSaviq, In HUD and dbustest we're keeping them separate and managing the dev packages appropriately.20:31
Saviqtedg, right, that's what I thought, so whatever ends up in the _package_ name is the API version20:32
tedgSaviq, So then when you update to libhudclient2 you change your builddep20:32
tedgSaviq, Of the dev package name is the API version20:32
Saviqtedg, and the lib package name?20:34
tedgSaviq, That will be the ABI version.  But, really, you don't care about that as dpkg discovers the right version for you.20:35
Saviqtedg, yeah I know, just trying to get the hang of it20:35
Saviqtedg, so libhudclient2-dev might very well relate to libhudclient50?20:36
Saviqor the other way round, rather20:36
tedgWell, they could be either way really :-)20:36
tedgIn theory you'd always change ABI when you change API.20:37
* tedg tries to think of a way you could change API without changing ABI...20:37
tedgI guess if you changed #define values or enums.20:37
=== francisco is now known as Guest87879
=== salem_ is now known as _salem
slangasekbregma: hey, so if https://code.launchpad.net/~vorlon/compiz/lp.763148-0.9.9/+merge/157584 is merged onto lp:compiz/0.9.9, do you know when I should see this show up in raring?22:48
bregmaslangasek, probably tomorrow, after the autolander has run22:55
bregmaI'm not sure what time they schedule that for exactly22:56
slangasekbregma: ok, that was going to be my next question ;)  since nearly a full day has passed since it was merged already22:56

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