=== duflu_ is now known as duflu [05:25] karni: oh, right, thanks (regarding savi_q) [08:30] MacSlow: does http://s-jenkins.ubuntu-ci:8080/job/autopilot-testrunner-otto-trusty/2242/testReport/junit/unity8.shell.tests.test_notifications/InteractiveNotificationBase/test_sd_incoming_call_Desktop_Nexus_4_/ make any sense to you? [08:56] jamesh, thostr_: morning! https://bugs.launchpad.net/unity-scope-mediascanner/+bug/1272218 [08:56] Ubuntu bug 1272218 in Unity Media Scanner Scope "Latest lp:unity-scope-mediascanner scope does not work" [High,New] [08:57] sil2100: hi. I discovered a problem while testing the scope today: I missed one location when fixing the scope for the API renaming [08:57] sil2100: trunk is now working with the new scopes API version of unity-scope-tool for me [09:06] jamesh: will it now work even with the old scopes API? [09:26] tsdgeos, hm... looks like the tap/click on the button didn't happen. [09:27] MacSlow: can you try seeing if you can repro it? it's the only failure in autopiloot we are having now (now that otto decided not to crash every other run again) [09:30] tsdgeos, I'm in NotifyOSD bug-fixing land atm... I'll give it a deeper look laster today. [09:30] oki === mpt_ is now known as mpt [09:34] sil2100: sorry, was finishing off a meeting. If I branch the mediascanner scope from before the updates to the new API, how difficult would it be to have packages for both? [10:28] sil2100: so, the lp:unity-scope-mediascanner/saucy branch looks like what we'd want to parallel install with the new binding. Would updating the package name of the new scope to unity-scope-mediascanner2 be enough? [10:47] jamesh: hmmm, I think that maybe we should just wait for the switch to the new scopes API in overall? [10:48] well, we want to be able to test out the new code. I know thostr_ is very keen to see this land, since it has taken longer than expected. === _salem is now known as salem_ === MacSlow is now known as MacSlow|lunch [12:20] Cimi: if the test launches an app etc, I'd put it into the app lifecycle tests [12:23] ok [12:23] mzanetti, or upstart?+ [12:24] Cimi: not sure why the app lifecycle and upstart tests are not groupes [12:24] grouped [12:24] Cimi: ah ok. no. thats different [12:25] Cimi: test_upstart tests if unity8 communicates correctly with upstart when unity8 is starting [12:25] so app lyfecycle [12:25] lifecycle [12:25] Cimi: the app lifecycle tests are the ones, yeah [12:25] ok [12:27] mzanetti, same check in upstart file works, not in app lifecycle [12:27] I'm debugging [12:27] with works I mean, it fails when it is expected to fail [12:27] Cimi: hmm... dunno... actually imho the lifecycle tests should be in the unity8 directory too [12:28] not sure why we have a separate directory for that [12:28] let me create a separate dir [12:28] dir/file [12:32] where is the unity8 default/user configuration stored? [12:32] ups [12:32] default *launcher* configuration [12:47] mzanetti, ^ [12:48] seb128, I think in the source code, might be wrong [12:48] seb128: in dconf [12:48] mzanetti, where is the schemas? is that the one from unity7? [12:48] seb128: yes [12:49] mzanetti, thanks [12:49] seb128: not that this is only the default config [12:49] seb128: if you change the launchers config it won't be stored in there any more [12:49] mzanetti, accountsservice then right? [12:49] seb128: yep [12:49] mzanetti, how do I reset it? [12:50] seb128: gdbus call --system --dest org.freedesktop.Accounts --object-path /org/freedesktop/Accounts/User32011 --method org.freedesktop.DBus.Properties.Set com.canonical.unity.AccountsService launcher-items "<[{'defaults' : }]>" [12:50] mzanetti, in fact my question is "is ubuntu-system-settings in the default launcher config nowadays" [12:50] mzanetti, let me try that, thanks === MacSlow|lunch is now known as MacSlow === dandrader is now known as dandrader|afk [13:13] tsdgeos: "fun" fact. while qml/Components/FilterGrid.qml and tests/qmltests/Components/tst_FilterGrid.qml are same in my branch (any seen changes would not affect that behavior), if you set maximumNumberOfColumns on the FilterGrid to 2 instead of 3 (tests contain value of 3), trunk code shows them in 2 columns, while new-scopes shows them in... *1* column, even though it should show *2* columns. [13:13] tsdgeos: I yet have to understand why that happens. testFilterGrid with maximumNumberOfColumns set to 2 in tst_FilterGrid.qml in unity trunk and new-scopes trunk shows the problem. [13:15] tsdgeos: oh, sorry: http://paste.ubuntu.com/6808249/ [13:16] These changes would not affect the behavior of FilterGrid for models with the test default number of items. And I did try to manually set "1" for collapsedRowCount, did not help. === alan_g is now known as alan_g|lunch [13:21] tsdgeos: I guess that would mean my previous change to ResponsiveGridView spacing/margins could break that. There's no other resonable reason. [13:21] * karni reviews code [13:35] tsdgeos: today's recipe got a bit easier and I might have missed the rm -rf from yesterday's: http://pastebin.ubuntu.com/6808352/ [13:37] tsdgeos: correction, http://pastebin.ubuntu.com/6808355/ === dandrader|afk is now known as dandrader [13:38] tsdgeos: and I got http://pastebin.ubuntu.com/6808345/ [13:39] Yes. With maximumNumberOfColumns set to 2, it looks like ResponsiveGridView changes are to blame. [14:01] karni: i got lost to be honest :D [14:02] tsdgeos: Looks like grid spacing fixes badly affect number of columns when maximumNumberOfColums < 3 (specifically, 2). [14:03] tsdgeos: I'm looking into it, I just wanted to share. I found out more while I was messaging you. Sorry for the confusion. [14:03] I must have sounded bit random. [14:03] karni: no worries, good you're on track :-) [14:03] :) [14:07] tsdgeos: I have one question though. This snippet [1] comes from ResponsiveGridView. If the last column does not contain spacing to it's right, can you tell me why would we be dividing by (delegateWidth + spacing)? [1] http://paste.ubuntu.com/6808495/ [14:08] karni: because it's probably having spacing to its right [14:08] and yes that may be a difference to how the vJournal is coded [14:08] tsdgeos: I talked to Katie. If all columns have spacing to it's right, and same margin on both sides, then the free space at the right edge of the screen is larger than at left side. [14:09] tsdgeos: I think that formula is fundamentally wrong, even if the rendering was right. Following that, spacingForColumns() would also require adjustment. [14:09] yes [14:09] tsdgeos: I showed her two screenshots, with and without my fix [14:09] the formula may be wrong [14:09] but it's rendering right [14:09] if you fix the formula [14:09] is rendering wrong [14:09] so you'll need to fix the other stuff too [14:10] tsdgeos: Right. I was just asking for sanity check if you consider the formula *could* be wrong. I think it is. [14:10] let me read it [14:10] Sure. qml/Components/ResponsiveGridView.qml [14:10] columnsForSpacing function [14:12] tsdgeos: Limiting our conversation for sake of sanity to this function only, that's what I suggest: http://paste.ubuntu.com/6808514/ [14:14] tsdgeos: basically, columnsForSpacing should return: (parent.width - bothMargins - delegateWidth) / (delegateWidth+spacing) + 1 (the last column without spacing) [14:15] Been a while I tried to convey mathematical reasoning over IRC. [14:17] he he [14:17] Do you think what I said makes sense? [14:17] doing my math [14:17] great === alan_g|lunch is now known as alan_g [14:21] tsdgeos: I just sent you a drawing ;) [14:22] karni: (parent.width) / (delegateWidth + spacing) makes more sense to me [14:22] but let me read your drawing [14:22] tsdgeos: I agree [14:23] i wonder who originally did that code [14:24] tsdgeos: bzr blame qml/Components/ResponsiveGridView.qml (I dislike the command, FWIW) [14:24] karni: nah that's a lie [14:24] we moved the code [14:24] tsdgeos: Michal, but shows you could have touched that line. [14:24] so it's going to say saviq [14:24] but he did not do it [14:24] I see. I don't care :) as long as we can understand it. [14:25] gerry is more acurate [14:25] http://bazaar.launchpad.net/~unity-team/unity/phablet/changes?filter_file_id=responsivegrid.qml-20121121162541-qc48wqyt6pxgw0ib-1 [14:25] columnsForSpacing has been the same since revision1 [14:25] http://bazaar.launchpad.net/~unity-team/unity/phablet/revision/106 [14:25] yes [14:25] I just looked, you're right. [14:25] i'd say that what i said makes more sense that what gerry coded [14:25] but then he has a math phd [14:26] so i won't say he's wrong :D [14:26] hahahahah [14:26] You made me laugh.. thank you. I had some hard time recently heh. [14:26] :) [14:26] :) [14:26] what i suggest still works on trunk [14:26] so if it helps with your usecase [14:27] yes, thank you [14:27] i don't see any problem in getting it merged [14:27] let me see if that helps [14:27] we could even probably find out a width for the test that it breaks [14:27] in current trunk [14:28] Yes. I want to check if that works for 2 columns though. [14:28] tsdgeos: but it's a good finding. our both formulas where equivalent, yours was cleaner. [14:29] (using the fact that spacing = 2*margin) [14:31] tsdgeos: wohoo it works for 2 columns. Now I'll get the spacing sorted, and we should be good. [14:31] nice === ricmm_ is now known as ricmm === rachelliu_ is now known as rachelliu [14:59] karni: that columns fix, you're doing it over trunk or over new-scopes? [15:00] tsdgeos: I am over new-scopes, but I understand it applies to trunk, so once I have it ready, we can do it over trunk :) Don't want to be moving the party over in the middle. [15:00] oki [15:16] tsdgeos: So, yeah.. I'm actually working with trunk already. If you $ tryFilterGrid, you'll see these 3 columns are not center aligned. [15:17] And that's because margins always evaluate to zero width, because "allocatable horizontal space" is spread over column spacings [15:17] right === alan_g is now known as alan_g|tea === dandrader is now known as dandrader|lunch [15:48] Mirv: those upgrade instructions work better [15:48] Mirv: and yes, stuff is asserting in the qtubuntu QPA plugin :/ === alan_g|tea is now known as alan_g [16:14] Mirv: after recompiling qtubuntu all works for me [16:15] Mirv: which is weird since you seem to recompile it :_S [16:18] well, all works until you blank the screen [16:18] that is [16:18] :D [16:18] then you can't unblank it anymore === dandrader|lunch is now known as dandrader [16:59] mhr3: Hey man, you around? Do you think you could check my math? [17:00] karni, only if you use integrals [17:00] karni, or at least some derivations [17:00] mhr3: I wish ;) http://paste.ubuntu.com/6809314/ [17:00] mhr3: I had to comment the code for sanity [17:01] pff, addition and division.. boring :P [17:01] I know, right.. [17:02] mhr3: problem is, margins are still too large. if you set maximumNumberOfColumns to 2, it's always 1 column. problem does not show for 3+ columns. [17:03] literally, if you set it to 2 (and ResponsiveGridView has leftMargin and rightMargin = margin/2 (as in code)), it renders stuff in one column. [17:03] The math is sound to me. [17:04] so what's the inputs? [17:04] mhr3: for the record, I confirmed with Albert that formula for columnsForSpacing() is correct [17:05] mhr3: maximumNumberOfColumns. [17:05] mhr3: for 3 and above, it's correct. for value of 2, it becomes 1 column instead. [17:07] mhr3: I assume margins for a GridView define the padding on the outside of the view? [17:07] actually, that wouldn't even manner. parent.width doesn't change, the total width should match anyway. [17:09] karni, could you console.log all the params for columnsForSpacing when max == 2? [17:10] mhr3: in onCompleted of the gridview? [17:10] karni, just print it every time the func is called [17:10] ah sure [17:13] preferably when there aren't dozens of categories :) [17:15] mhr3: spacing 8 parent.width 240 delegateWidth 48 spacing 8 columns 4 - but they're limited by Math.max(4, maxvalue=2) [17:15] I also know the number of columns is 2, because when filtered to one collapsedRowCount (even though showing in 1 column), there's 2 items. [17:16] So, the number of columns is 2, and it is correct that max of 4 columns would fit. [17:16] let me do the same with spacing, which is where the problem seems to be [17:17] now i lost track of what is actually wrong... [17:17] cause it sounds like it works how it's supposed to [17:17] mhr3: collapsedRowCount: 1, maximumNumberOfColumns: 2 ---> results in single column of 2 items. [17:18] if I manually lower the margins (that's where the math is), these two items fit in one row. [17:18] Tests didn't catch it because the formula for column count was wrong, and tests are implemented for 3 columns. [17:18] mhr3: Anyway, I'll print stuff out and see if it adds up. [17:19] but collapsedRowCount doesn't come into play here at all [17:19] mhr3: but ResponsiveGridView is part of FilterGrid, and testFilterGrid shows that problem. [17:20] The formula for calculating column count was fundamentally wrong. [17:20] It rendered fine, because the spacing also had a wrong formula. [17:20] all columns had spacing (and the last one shouldn't), that's why stuff was left-aligned [17:20] if you tryFilterGrid in unity8 trunk, you'll notice they're left aligned [17:21] That's because margins always evaluate to zero width, and last column of the grid also has spacing. [17:23] FilterGrid should work independently of ResponsiveGridView margins, and it does not :) [17:24] i don't get this: [17:24] + // Margin is half of the spacing, thus L + R margin = spacing. [17:24] + return Math.max(1, Math.floor(parent.width / (delegateWidth + spacing))); [17:25] you say that margins == spacing, but it's not accounted for in the formula [17:25] mhr3: check your mail for picture [17:25] didn't get anythin [17:25] g [17:26] then check now :) [17:26] mhr3: it is, because far right delegateWidth+spacing is actually delegateWidth+(2 x (spacing/2) ) [17:26] where spacing/2 = one margin [17:26] That's where the comment came from. [17:26] ah, right [17:26] a sec, need to find a pen [17:27] sure :) [17:27] and some paper :) [17:27] mhr3: I enjoyed drawing a little. [17:27] btw does parent.width include the margins? [17:29] mhr3: yes, in my understanding. [17:30] mhr3: if I do: left/rightMargin: margin/2.4 (instead of margin/2), it works fine. which means, sum of both margins is still to large :/ [17:30] or I'm missing a detail in the layout [17:30] parent.width is the width we have available, I assume that means: [ Lmargin | grid grid grid | Rmargin ] [17:33] doesn't that just mean that spacing != 2*margin [17:33] ? [17:33] wait. let me set left margin 0, right margin = spacing. if it fits, the spacing/2 should as well. [17:34] pff.. no, stil 2 items don't fit in one row, although visually they would, with large spacing. [17:35] mhr3: I need to grab lunch (lol), back in a sec, in case you'll still here. If not, no worries, catch you on Monday :) [17:35] hm, lunch, that's a good idea... [17:51] mhr3: heh, I seem I'm not the only one eating at weird hours ;) [17:51] *it seems [17:53] i hate myself when i do this :P [17:54] mhr3: I think there's more of us in this company hahah [17:55] by the way, parent with 240 = 36 margin + delegate(48) + spacing(72) + delegate(48) + 36 margin --- margin := spacing/2.0 [17:55] wtf.. [17:57] something here has thick borders lol [17:59] I sense a problem around cellWidth: delegateWidth + horizontalSpacing [18:08] mhr3: Well. If the margins (as it is) has always evaluated to value 0, and it didn't bother anyone, I guess I should fix the code to just tell the truth, and assume there are no margins. That's already how tryResponsiveGridView renders. [18:27] karni, well if cellWidth is calculated that way, margin has to be 0 indeed [18:47] mhr3: I think I need to consult this with design. Margins make little sense there. [18:47] mhr3: In any case, I'll calling this a wrap. Have a good weekend, buddy :) === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [20:43] So I have this unity8 autopilot test that fails in jenkins but works for me on the device. How can I make my test env more like jenkins? [20:43] Saviq, mzanetti ^ ? === salem_ is now known as _salem === _salem is now known as salem_ === salem_ is now known as _salem