=== thomi is now known as thomi|lunch === shiznix_ is now known as shiznix === thomi|lunch is now known as thomi [06:52] Wellark, ping [07:14] sil2100, didrocks, can we just push https://code.launchpad.net/~mhr3/unity/fix-ap-preview-tests/+merge/175123 directly? [07:14] * mhr3 is really annoyed with unity unit tests [07:15] mhr3: +1 [07:18] didrocks, sil2100, pushed to lp:unity [07:19] \o/ [07:20] hm, today I started my system and right after logging in I got like 10 apport error dialogs [07:20] huh === alan_g|EOD is now known as alan_g [08:08] Saviq, is it just my impression or is Jenkins performing really really badly lately? [08:08] dandrader, it is [08:08] dandrader, we're killing the pandas, it seems [08:09] dandrader, most probable - memory overheating [08:09] wow [08:09] dandrader, and then there was a licensecheck script failure [08:09] dandrader, but yeah, https://jenkins.qa.ubuntu.com/job/unity8-ci/ looks tragic :/ [08:10] dandrader, they're looking to throttle the builds - slower, but at least (potentially) completing [08:11] dandrader, and also the calxeda setup finally arrived - ETA on it is around a month, though [08:11] ah, cool [08:17] mhr3: hi! Are you the one currently doing unity fixes? Since I noticed, through the indicator stack, that one of the AP tests started failing [08:18] sil2100, hey, no [08:18] i fix my stuff [08:18] Too bad ;) [08:46] Cimi: ping [08:47] dednick, he should be away today [08:47] Saviq: ah. thanks [08:53] dednick, :P [08:54] huh [08:55] ? [08:55] dednick, re: semi-colons [08:55] how is it that PS Jenkins Bot already commented PASSED on some MRs [08:55] but http://s-jenkins:8080/job/unity8-ci/ doesn't yet know? [08:56] ah, archiving artifacts [08:57] Saviq: they're finished now [08:57] Saviq: oh. hm, no they're not [08:58] that's not a very good ci track record. 2 green in 30 attempts... [08:58] dednick, yeah [08:58] dednick, pandas are dying [08:58] dednick, memory overheating, apparently [08:59] lol. [09:00] i think there's something up with saucy. My machine has been dying the last week or 2. [09:01] dandrader, you approve the addrenderer branch *a lot*, heh? ;) [09:02] geh, unity AP testing blocks the whole check infrastructure ;/ [09:03] Saviq, got anything I can look at? Busy waiting on reviews. [09:04] nic-doffay, busy waiting? ;) [09:04] Saviq, not busy waiting :P [09:04] nic-doffay, start implementing the filters panel [09:04] nic-doffay, as seen in http://design.ubuntu.com/apps/building-blocks/option-selector [09:05] my nexus is buggered. wont turn on. [09:06] dednick, tried putting it on power for some time? [09:06] dednick, and is that galaxy or 10? [09:06] Saviq: pong [09:06] nic-doffay, you should be able to work on top of https://code.launchpad.net/~unity-team/unity8/filters/+merge/173691 [09:06] Saviq: yeah, powered, galaxy [09:06] dednick, disconnect, remove battery, replace battery, try turning on disconnected [09:07] Wellark, did you hear/know about the hud getting stuck somewhat? [09:07] Wellark, that's how we encountered the hud timeout in our HudClient plugin [09:07] Wellark, and it's been happening to different people for some days now [09:07] Wellark, hud-service is apparently running [09:07] Wellark, but DBus to it times out [09:08] Wellark, and unity7 hud is non-responsive (i.e. just sits there doing nothing when you type) [09:08] Saviq, I'll look into it. [09:08] Saviq: yay! it's back! [09:08] Saviq, what still needs to be done with that branch before it's merged? [09:09] dednick, cool [09:09] dednick, it sometimes gets confused when on power [09:09] nic-doffay, I need to re-review it [09:09] Saviq: ta [09:10] nic-doffay, should be merged this week [09:10] Saviq, k [09:10] Saviq: have people updated to the latest saucy packages and restarted their session? [09:10] Saviq: at some point session bus started to be controlled by upstart [09:10] Wellark, I did, not seing it now [09:10] Saviq: so at least for me I was seeing this dbus timeout [09:11] Wellark, so if you think that fixed [09:11] that was because there were two session bus daemons runing [09:11] and hud was using the upstart one [09:11] Wellark, then that's ok, will report back if we encounter again [09:11] and rest of the desktop was using the old one [09:11] Wellark, got it [09:11] Saviq: thanks! [09:11] Wellark, cheers [09:17] Saviq, how would someone navigate to these filters? [09:17] nic-doffay, the "grape" icon [09:17] nic-doffay, it's a button on the search entry [09:17] Saviq, a grape?! [09:17] nic-doffay, the dots [09:17] nic-doffay, to the right of the search etnry [09:17] Saviq, ah [09:17] entry [09:24] dednick, I'm on holiday :-P [09:24] dednick, tell me [09:24] Saviq, Cimi: https://code.launchpad.net/~macslow/unity8/fix-button-tint-hint/+merge/175314/comments/393708 [09:24] I'll be here for a while, upgrading the phone [09:25] Cimi, do you know when Jouni is back from vacation? [09:25] MacSlow, I think today or tomo [09:25] Cimi: dont worry about it [09:25] dednick, well I'm here [09:25] Cimi: you back tomorrow? [09:25] dednick, nope, monday [09:25] dednick, going to Milan for Depeche Mode concert :-P [09:25] Cimi: awesome [09:26] Cimi: need you to fill in some status of settings/indicator components at some point. https://docs.google.com/a/canonical.com/spreadsheet/ccc?key=0Al0gMJhDnGQEdHA1S2VRNmhRdHFvVFNpZ1k5VlA5NEE#gid=1 [09:26] MacSlow, on your comment [09:26] MacSlow, which theme are you using? [09:26] Cimi: the ubuntu-settings-components section. [09:26] MacSlow, with SuruGradient the colour should be what you want [09:26] MacSlow, cheers [09:27] Cimi, "Suru"? [09:27] nic-doffay, review fail https://code.launchpad.net/~macslow/unity8/fix-button-tint-hint/+merge/175314/comments/393708 [09:27] MacSlow, you should use SuruGradient [09:27] MacSlow, the shell sets the theme in Component.onCompleted on shell.qml [09:29] Saviq, egh [09:29] nic-doffay, actually maybe not ;) [09:29] MacSlow, in your qmltest set the theme with SuruGradient [09:29] Cimi, but using a gradient for the text-color of summary and body doesn't make sense [09:29] MacSlow, the palette is the same of SuruDark [09:30] mostly [09:30] Cimi, the text-color is a color not not a gradient :) [09:30] it's the fuckin name of the theme :) [09:30] MacSlow, it's just a theme name ;) [09:30] MacSlow, look at shell.qml [09:30] Component.onCompleted [09:30] calm down peopl ;) [09:30] ahah I am joking :) [09:31] MacSlow, http://pastebin.ubuntu.com/5886983/ [09:32] +1 [09:32] Saviq, which highlights that the notification are not using bg colour from the palette [09:32] Saviq, otherwise the background would have changed accordingly to the theme [09:33] MacSlow, Cimi http://ubuntuone.com/6nL2LHe0VX3yyXETQjA5CJ [09:34] mhr3: unity-applicati invoked oom-killer <- doesn't sound too good I suppose? [09:34] Cimi, that's your failure, isn't it ;) [09:34] sil2100, say what now? [09:34] Cimi, MacSlow that's "SuruGradient, old, Suru" in there [09:35] mhr3: I'm looking at unity AP testing for ati and this popped up, and anyway the whole machine takes too long to finish - you saw something like that before? [09:35] jibel: ^ ? [09:35] jibel: maybe it's the OOM issue you mentioned before? [09:35] http://10.97.0.1:8080/job/autopilot-saucy-daily_release/label=autopilot-ati/484/console [09:35] sil2100, isn't it just killed because something else wants more mem? [09:35] Cimi, Saviq: I didn't know about the required changes needed to be done to Notifications... I only know about the set of colors I once got from Jouni [09:36] MacSlow, yeah, they're the same in SuruGradient theme [09:37] sil2100, that's something else [09:37] Saviq, what env-var to set to force a particular theme for the test? [09:37] Saviq, or did you hack that temp. into the test? [09:37] MacSlow, http://ubuntuone.com/6nL2LHe0VX3yyXETQjA5CJ [09:37] jibel: this machine seems hanged, no tests are being ran, I'm thinking of killing it [09:38] * didrocks things jibel killed compiz :p [09:38] thinks* [09:38] sil2100, I mean it is OOM but not caused by RMD as it used to be [09:39] I would like this machine to finally finished, since there are like 5-6 check jobs queued ;p [09:39] jibel: you think we can abort it? [09:40] Oh, it's moving forward [09:40] jibel: but there was a kernel stack trace :D [09:42] sil2100, the kernel call trace says there is a page fault with is why OOM killer has been invoked [09:42] sil2100, and the culprit seems to be unity-applicati [09:43] MacSlow, you can see the colours here http://bazaar.launchpad.net/~ubuntu-sdk-team/ubuntu-ui-toolkit/trunk/view/head:/modules/Ubuntu/Components/Themes/SuruGradient/Palette.qml [09:43] yay for very useful short names [09:43] Saviq, not really [09:43] Saviq, because the black is not the bg colour of surudark/gradient [09:43] Cimi, ;) [09:43] Saviq, ideally we should create a theme for the notifications [09:43] Saviq, and set the theme for them [09:44] Cimi, but we can't [09:44] Saviq, with SuruGrdient as parent theme [09:44] Saviq, not yet, one theme for each qmlscene currently [09:44] yup [09:44] Saviq, same thing for the indicators [09:44] Cimi, yeah [09:45] is this good to go? https://code.launchpad.net/~unity-team/unity8/unity8.background_gsettings/+merge/174958 [09:45] sil2100, checking if there's a leak somewhere [09:45] Cimi, no-one from us reviewed it yet [09:45] ok [09:46] hmm, maybe the diff models cause it [09:47] Saviq, I believe we can use CrossFadeImage from sdk [09:47] Cimi, yeah, we should [09:47] Saviq, separate branch pls [09:48] Saviq, I can do that on monday [09:48] Cimi, k [09:48] Saviq, I haven't built unity8 in a while. Missing two packages. libusermetricsoutput-1 and unity-shell-launcher=2 any idea where I can find them? [09:49] nic-doffay, ./build -s; ./build -c [09:50] nic-doffay, you should start from trunk with filters merged, though [09:50] nic-doffay, as filters might not have the ./build fix yet [09:50] Saviq, ah right. [09:51] Cimi, Saviq: are there any other required changes needed to happen to notifications due to the introduced themeing? [09:51] MacSlow, we'll need to look with designers to see if the notifications background should be changed [09:51] MacSlow, if not, nothing we can do now [09:51] MacSlow, as we'd need a different palette for notifications [09:52] MacSlow, but we can only use a single theme for the whole QML tree [09:52] Saviq, ok... I guess we'll have to wait for Jouni to come back. I updated my MP. [09:52] MacSlow, he's back today, actually [09:53] dednick, any ideas on why this might be happening? https://jenkins.qa.ubuntu.com/job/generic-mediumtests-runner-saucy/1051/testReport/junit/unity8.indicators_client.tests.test_battery/TestDisplayMenus/test_brightness_slider_with_mouse_/ [09:55] dandrader: when you run your shell app does your battery indicator have menu items? [09:55] dednick, on the desktop? [09:55] dandrader: ya [09:56] dandrader: ah. crap. that's jenkins. hm [09:57] jibel: how long does stopping the container take? [09:59] dandrader: i'm not sure. something is going on with the page not loading. [10:00] sil2100, depends on the size of the delta, which is 16G for the last run [10:00] * sil2100 is waiting impatiently [10:00] Yay [10:01] But there was a timeout anyway [10:04] dednick, on my desktop it fails but for a completely different reason: http://paste.ubuntu.com/5887039/ [10:04] dednick, how am I supposed to run this test? [10:04] * dandrader loves autopilot [10:05] dandrader, it's not really like it's autopilot's fault ;) [10:05] dandrader: looks like the same reason to me [10:06] dandrader, cd tests/autopilot; PYTHONPATH=.:$PYTHONPATH autopilot run unity8.indicators_client.tests.test_battery.TestDisplayMenus.test_brightness_slider [10:07] Saviq, same result. does it work for you? [10:07] dandrader, yes [10:07] dandrader, I do have indicator-power installed, do you? [10:07] dandrader, indicator-battery, even [10:07] indicator-battery [10:08] works for me as well [10:08] dandrader: which branch is it? [10:09] dednick, on Jenkins it seems QDbusActionGroup, from QMenuModel, is missing a actionAppear signal [10:09] dednick, on my desktop it doesn't complain about it [10:09] dandrader: yeah, i've checked the code though, its there. [10:09] dandrader: qt doesnt output on autopilot. [10:09] sil2100, yea, there seems to be bug in the diffing algo, perhaps we should disable it so we finally get a unity release? [10:09] dednick, https://code.launchpad.net/~dandrader/unity8/lp1116207/+merge/175163 [10:10] dednick, yeah it does, if the test fails [10:10] dednick, https://jenkins.qa.ubuntu.com/job/generic-mediumtests-runner-saucy/1051/testReport/junit/unity8.indicators_client.tests.test_battery/TestDisplayMenus/test_brightness_slider_with_mouse_/ [10:10] dandrader: 113 +add_subdirectory(QMenuModel) [10:10] you're mocking qmenumodel [10:11] Saviq, I do have indicator-power [10:11] dandrader, indicator-battery [10:11] dandrader, sorry [10:11] Saviq, "Package indicator-battery is not available, but is referred to by another package." [10:12] dandrader, yeah, I probably have it from ppa:phablet-team/desktop-deps [10:13] dednick, yes. But I wonder how come autopilot would use mock plugins instead of the real ones [10:13] dednick, just to check it I've added that signal to my mock impl + some deb prints to stderr but nothing changed... [10:13] regarding the autopilot error, I mean ^ [10:14] Saviq, so I should have indicator-battery installed in my desktop anyway? [10:14] mhr3: diffing algorithm? [10:14] dandrader, you can [10:14] dandrader, it's later going to be replaced by indicator-power [10:14] dandrader, and for that test - yes, you need to [10:18] Saviq, indicator-battery installed. autopilot tests still fails :/ [10:19] dandrader, ./builddir/src/Panel/Indicators/client/indicators-client [10:19] dandrader, can you see "Battery"? [10:19] dandrader, if so, when you open it, can drag the slider about / switch the checkbox? [10:20] Sarvatt, there's a "Battery" option, but when selected I just get a "Empty" text [10:21] sil2100, feature that probably caused the oom, here's the branch to disable it - https://code.launchpad.net/~mhr3/libunity/disable-diff-models-until-fixed/+merge/175522 [10:22] mhr3: are there any side-effects of disabling that? [10:22] sil2100, yea, no crashes :) [10:23] ;p [10:23] otherwise, not really [10:25] Saviq, indicator-battery installs a indicator-battery-service binary that segfaults when I try run it manually. does it work for you? [10:25] dandrader, yeah [10:26] s**t [10:26] dandrader, 12.10.6phablet1bzr12raring0 here? [10:26] Saviq, yes [10:27] dandrader, try the one from ppa:phablet-team [10:27] dandrader, https://launchpad.net/~phablet-team/+archive/ppa/+sourcepub/3354110/+listing-archive-extra [10:29] dandrader, there was a bug on it (that dednick found through autopiloting indicators_client) [10:29] dandrader, you're probably hitting the same [10:32] Saviq, yeah, now it doesn't crash anymore [10:32] dandrader, cool [10:33] ok, no I see something with that indicators-client demo binary [10:33] s/no/now [10:36] wtf was all that. restarted machine and all hell broke loose [10:36] ...sigh. partial upgrade [10:39] hmm... that battery-brightness test worked fine locally [10:40] dandrader: missing indicator-battery? [10:40] dednick, yes [10:40] dandrader: wonder why it wasnt working on jenkins. [10:40] as in couldnt find the qmenumodel [10:41] dednick, my last desperate debug attempt was to remove my qmenumodel mock. let's see what jenkins says now... [10:42] if it still complains then I will consider it "random jenkins failure" and click rebuild till the end of times === dandrader is now known as dandrader|afk [10:48] dandrader|afk: why are you installing your mock? [10:56] jeez we're down to 3 pandas :/ [10:56] no wonder everything takes so long === dandrader|afk is now known as dandrader [11:03] dednick, all mocks get installed [11:03] dednick, because jenkins runs all tests from packages [11:04] dandrader: i think it's only ones needed for autopilot tests. other tests are run from showdow build no? [11:04] so if mocks are not installed, qmluitests in jenkins won't be able to find them === MacSlow is now known as MacSlow|lunch [11:04] s/showdow/shadow [11:06] dednick, that might be, I'm not 100% sure. I'll try that next [11:06] dandrader: as far as i know, only lightdm, application & metrics mocks are installed. [11:07] dandrader: but anyway, it shouldnt load up the mock qmenumodel first... [11:07] dednick, another odd thing is that I later implemented that signal in mock and jenkins still complained [11:08] dednick, dandrader autopilot is run from packages [11:08] dednick, dandrader, qmluitests are run in build tree [11:09] dednick, dandrader, correct - unity8-fake-env is what's used in autopilot for mocks (while we don't have the real ones) [11:10] dandrader: i wonder if the usr/lib/arch/qt5/qml import is added as a special case on the end of the imports, therefore getting last priority. [11:14] but then it probably wouldnt work on our machines. === shiznix_ is now known as shiznix === greyback is now known as greyback|lunch [11:22] dednick, it is added as fallback [11:22] dednick, if you're running installed [11:22] Saviq: yeah, which should be last [11:22] dednick, actually no, our *private* dir is added as fallback [11:22] dednick, *private*/mocks overrides [11:23] or does it [11:23] unity8/qml/mocks rights? [11:23] * Saviq got lost again [11:23] well, both unity8/qml & unity8/qml/mocks [11:23] dednick, yeah, system-wide imports get precedence [11:24] dednick, then, our private unity8/qml, then mocks [11:25] dednick, you could argue that it should be: unity8/qml, qt5/qml, unity8/qml/mocks [11:26] I'm not convinced either way, tbh [11:26] maybe we should even have separate dir for overrides [11:26] Saviq: shouldnt really matter. we shouldnt have unity8/qml that overrides qt5/qml [11:26] dednick, I agree [11:26] so unity8/qml/override, qt5/qml, unity8/qml, unity8/qml/mocks [11:26] but we should avoid that if possible [11:30] Saviq, dots to the right of the search entry? We're talking about the search tab on the left of the indicators, correct? [11:36] nic-doffay, no [11:36] nic-doffay, tap on it in Apps, for example [11:36] nic-doffay, search entry shows up [11:36] nic-doffay, the button with dots should show be at the right of the search etnry [11:38] dandrader, thanks for taking on some reviews, btw! [11:39] Saviq, I've nothing else to do :) [11:39] Saviq, no button there. I'll ask Pawel. [11:39] nic-doffay, Paweł is on holidays [11:39] Saviq, oh dear. [11:39] Well the search entry appears. [11:39] nic-doffay, and indeed that is where *you* need to put he button ;) [11:39] Saviq, ah I assumed there was a button already! [11:40] nic-doffay, ass-u-me [11:40] Saviq, :P [11:40] ;) [11:40] Saviq, any word on assets or who I should speak to about them? [11:40] nic-doffay, jounih [11:40] nic-doffay, but check that they're not there somewhere [11:40] Saviq, who is also on holiday. [11:40] nic-doffay, sec, /me looks in the old repo [11:40] nic-doffay, no, back today === greyback|lunch is now known as greyback [11:42] nic-doffay, got it, sec [11:44] dandrader, we can always find something :) [11:44] Saviq, I'm all ears [11:44] dandrader, but reviews good, good :) [11:45] dandrader, bug #1193419 could be nice to have fixed [11:45] bug 1193419 in Unity 8 "Running apps should always be visible when navigating to apps lens" [Critical,Triaged] https://launchpad.net/bugs/1193419 [11:46] dandrader, and really, any of https://bugs.launchpad.net/unity8 [11:46] dandrader, or https://docs.google.com/a/canonical.com/spreadsheet/ccc?key=0AoGvOYxwuvpFdEJ5dURFb3Y0cnlKeEcxc0piNDZrWXc#gid=0 [11:46] dandrader, should be good enough busy-wait [11:46] Saviq, btw, how's that running vs. recent apps [11:47] Saviq, are we going to drop those thumbnails in the dash to close apps? [11:47] or switch to something else? [11:49] dandrader, the thumbnails will remain there, afaik [11:49] dandrader, but they're not really going to be running apps, recent instead [11:49] dandrader, whether we're going to allow closing / removing them is an open question still, AFAIK [11:52] Saviq, what qml file am I interested in to add the filter icon? [11:53] nic-doffay, PageHeader.qml [11:53] Saviq, ta === alan_g is now known as alan_g|lunch === dandrader is now known as dandrader|lunch === MacSlow|lunch is now known as MacSlow [12:45] Saviq, I see you also just started the CI for fix-button-tint-hint again [12:45] Saviq, I don't understand why it failed to merge and picked r116 instead of r117 [12:56] MacSlow, each CI run takes over 2hrs today [12:57] MacSlow, we're down to 3 working pandas :/ [12:57] fginther, hey is there anything we can do about that ↑? [12:59] Saviq, ping [12:59] Saviq, Not a whole lot. I can enable some of the less stable pandas and try to mitigate the problems there [12:59] olli_, hey, I won't be able to do the interview today, it's exactly at my EOD and I have to go out today [12:59] Saviq, that's perfectly fine [13:00] fginther, do we have a longer-term plan? [13:00] Saviq, outch [13:00] Saviq, yes, a new armhf build is in the works, [13:00] Saviq, we just have to find a way to make progress with our current hardware until then [13:01] Saviq, I am trying to move some builds of off armhf to reduce the build queues [13:01] but that's not a major help when the queue is full of unity8 [13:01] fginther, yeah, the "armhf build" you mention is the calxeda setup? so ETA ~1 month? [13:02] Saviq, right [13:02] Saviq, I've asked for other hardware for the short term, but no luck [13:03] fginther, ok well, I trust you'll do what you can for us [13:03] fginther, I'll try and reduce the load on there, too [13:03] we'll keep trying to make this better [13:03] fginther, by making sure we don't trigger unnecessarily [13:04] * Saviq needs to get on ang build the VM at long last [13:04] *and [13:04] so that we can run our UI / medium tests locally === dandrader|lunch is now known as dandrader === alan_g|lunch is now known as alan_g [13:31] nic-doffay, standup [13:41] kgunn, I have a pandaboard from previous desktop work. Can it be useful elsewhere? (I don't use it anymore) [13:42] fginther, ↑ [13:42] mterry: just about to ping you :) [13:42] mterry: yeah if fginther wants it great....otherwise, not much use [13:42] MacSlow, I've a minimal machine set up that we tried to work with with veebers [13:42] mterry: i heard a rumor those ti guysdecided to get out of mobile ;) [13:43] lol [13:43] MacSlow, but we were missing something and autopilot never ran in there [13:43] MacSlow, if you get somewhere, I'll ask you to dump your set of packages to see what's different [13:43] Saviq, d'accord [13:49] mterry: can you join #ubuntu-mir real quick [13:49] mterry: scratch that [13:52] fginther, could you use another pandaboard? [13:52] mterry, definitely, hang on (on the phone) === jono is now known as Guest30113 [13:58] Saviq, where I'm having problems with the InverseMouseArea conflicting with my new MouseArea in PageHeader. Any tips on how to get around this? [13:58] To do with selecting the filters. [13:59] nic-doffay, you just need to modify the dimension of the IMA, not sure what "conflicting" means there [13:59] Saviq, I'm setting the width. [14:00] nic-doffay, isn't width static? [14:16] Saviq, any ideas? http://paste.ubuntu.com/5887683/ [14:18] mhr3, hmm IIUC http://qt.gitorious.org/qt/qt/blobs/HEAD/src/corelib/tools/qlist.h#line288 [14:19] mhr3, ah no, 6 lines up [14:19] Saviq, yes, the assert is !isEmpty() [14:20] is that a bug in lvwph or is it because of the timer? [14:21] ie #5 vs #22 [14:21] mhr3, hmm it can be due to the timer... if it fired at a bad time [14:24] mhr3, but it would be a bug in lvwph nevertheless, it should protect itself from that, I'd say [14:24] i can hit this pretty much all the time with the diff models [14:25] now the question is, is it a bug in diffing or timer or the view :) [14:25] mhr3, try dropping the timer [14:25] mhr3, it's the easiest thing to do, probably, and would get you one step closer [14:26] Saviq, you mean to emit dataChanged on each update? [14:26] mhr3, yeah [14:26] mhr3, the count is only used to filter out empty categories [14:26] mhr3, so impact should be minimal with diffed models [14:27] mhr3, as we won't be doing 10 → 0 → 9 → 0 → 8 when you type [14:27] which means create → destroy → create → destroy → create [14:27]