[02:34] Hello, guys. [02:34] I am in some need of some serious help. [02:35] Anyone? [02:35] I'm having issues installing google chrome on Ubuntu. [02:37] ANYONE?! [02:39] Hello? === _salem is now known as salem_ === salem_ is now known as _salem [02:53] HELLOOOOOOOOOOOOOO [07:50] moin moin [08:00] eo [08:01] are there still promoted images? the last update on my dogfooding device came in on the 29th of august [08:02] mzanetti, what device/channel? [08:02] mako, ubuntu-touch/utopic [08:03] yeah, that's right [08:03] the google doc says the most recent promotion on utopic is #203 and it's almost 3 weeks old [08:04] yeah... that's why I asked if we gave up on those and only focus on the RTM branches for now [08:04] basically if I should switch the dogfooding device over to the other channel [08:08] mzanetti, no, we just need to get our shit together and promote something ;) [08:11] Saviq: did you then resubmit this one? https://code.launchpad.net/~mzanetti/unity/new-key-in-launcher-schema/+merge/232199 [08:11] mzanetti, no [08:12] mzanetti, see my comment [08:13] replied [08:13] Saviq: ^ [08:13] mzanetti, still, we need defaults in the schema, even if we do an override [08:13] mzanetti, why wouldn't we have defaults for the desktop [08:14] mzanetti, think unity8 desktop session [08:14] Saviq: so what should I put in there? [08:14] mzanetti, the same set we have in the override now [08:14] mzanetti, even more maybe, now that we can do clicks [08:14] the override has clicks too [08:14] does it? [08:14] yes [08:14] doubt it [08:14] we didn't support click updates [08:15] the defaults are not related to upgrades :) [08:15] ok maybe that got fixed in the mean time [08:15] meaning that we did not lose icons on upgrades any more [08:16] but we did have camera and gallery there before [08:16] yep [08:16] and removed them because they got broken on updates [08:16] ah... ok. didn't know we removed them [08:16] yeah, we only have non-clicks on launcher now [08:17] with your launcher rework, we're able to have clicks again, icons will get refreshed, right? [08:17] yes [08:17] ok we need to find out what we want on the launcher then, can you talk to Vesa? [08:17] ok [08:18] mzanetti, *and*, even if we wanted override, do you have a branch that moves the override to the new key? [08:18] not that we need the override any more [08:18] Saviq: well, I keep on reading the "favorites" key if the "items" key is empty [08:18] so yes, we could drop the override [08:19] mzanetti, I don't think that's necessary (to read favorites) [08:19] Saviq: works fine as an upgrade path from unity7 to unity8 though [08:19] mzanetti, yeah, *if* you leave the items empty [08:19] huh? [08:20] mzanetti, that would be a reply to my "why don't we have any defaults" question [08:20] ah, yes... if I put defaults in there it won't be empty [08:20] hmm... good point. [08:20] mzanetti, also, application:///, not application:// [08:21] mzanetti, I commented before [08:21] mzanetti, the problem there is that we can't use the favorites override to include appid:// entries can we [08:22] mzanetti, meaning we still can't have clicks there [08:22] meaning the upgrade path is a bit brokened :| [08:22] Saviq: yes, sure we can [08:23] mzanetti, well, except you'll get the new default set in your unity7 if you did not change anything [08:23] so you install the override and your unity7 launcher goes haywire [08:23] unity7 will just drop them [08:23] yeah, but it won't have firefox or anything either [08:23] yes, if you launch unity7 on a unity8 config it'll drop clicks. but keep the rest [08:24] mzanetti, no, what I mean is: [08:24] "the rest" is not useful in unity7 [08:24] why no? [08:24] because it's phone, contacts, messages, settings, webbrowser [08:24] yeah well. [08:24] on the desktop you won't have that override installed [08:25] and I expect contacts to become useful on the desktop [08:25] mzanetti, yeah, and that means your launcher is useless in your unity8 desktop session ;) [08:25] mzanetti, TBH I don't think it's worth it [08:25] Saviq: the whole uity8 desktop session is useless [08:25] Saviq: we don't have an applicationmanager [08:25] mzanetti, of course we do [08:26] in which case why would launching systemsettings be useless? [08:26] I don't get the problem [08:26] mzanetti, I'm not talking unity8 on X11, but unity8 on Mir [08:26] mzanetti, the problem is the two launcher configs are not compatible [08:26] that's why we created the new key [08:26] mzanetti, yes, and my point is [08:26] mzanetti, that we should not transition them [08:27] mzanetti, not by way of the other key [08:27] well... just tell me what to do then [08:27] mzanetti, just use items [08:27] mzanetti, don't read favorites, put the defaults for items in the schema, we can then drop the override [08:27] ok [08:28] mzanetti, we can have a task that will do a single-time copy of favorites (if non-default) to items [08:28] mzanetti, but again, I don't think it's worth it at this point [08:28] Saviq: what is the reason of the 3 slashes? [08:29] unity7 used 2 slashes only [08:29] mzanetti, uppercase [08:29] mzanetti, QUrl lowercases the hostname part [08:29] mzanetti, which is actually correct [08:29] mzanetti, so apps with differently cased .desktop files didn't work [08:29] hah... interesting [08:33] seb128: could you remind me where I can find the phone overrides for the launcher dconf keys? [08:33] mzanetti, I don't think we have phone overrides? [08:33] we do... quite sure about that [08:33] mzanetti, we have https://launchpad.net/ubuntu/+source/gsettings-ubuntu-touch-schemas [08:34] but those are schemas/defaults [08:34] mzanetti, ubuntu-touch-settings: /usr/share/glib-2.0/schemas/10_ubuntu-touch-settings.gschema.override [08:35] mzanetti, so right, in ubuntu-settings [08:35] seb128: thanks! [08:35] mzanetti, yw! [08:35] :) [09:01] Saviq, are we merging the big silo soon? [09:03] Cimi, it's blocked in proposed, currently trying to resolve this, but yes, it's almost there [09:04] Saviq, I would like to test flickable right speed on top of it [09:05] Saviq, I did some extra testing last night, and discovered few things [09:06] seb128: seems I can't push to ubuntu-settings. would you mind dropping the launcher overrides in there for me? [09:06] mzanetti, can't you mp your change? [09:07] hmm... maybe if I push to some +junk branch first [09:07] * mzanetti tries [09:07] oh [09:08] hmm.. no [09:08] mzanetti, don't bother, I though it had a proper vcs [09:08] the "Propose for merging is missing" [09:08] the "Propose for merging" is missing [09:08] yeah [09:08] it's hosted on a +junk vcs [09:08] Saviq, like the initial flick velocity is probably calculated wrong inside qt [09:08] mzanetti, what's the rational to drop the launcher config? [09:09] Cimi, lp:~ps-jenkins/unity8/ubuntu-utopic-proposed [09:09] Saviq, in any case I think the "1500" and "2500" (for deceleration and velocity) are not good default values ANYWAY, even scaled [09:09] seb128: the dconf + accounts service mix gave us a headache all the time, so we reworked that code bit. Now only using dconf, we added a new key for unity8 called "items" instead of "favorites" [09:09] Cimi, that's what's in proposed now [09:09] seb128: where we'll add the defaults just in there instead of an override [09:09] Saviq, dziekuje [09:09] mzanetti, did that land in utopic yet? [09:10] Saviq, so I think I will use a higher default value for now, and scale it [09:10] seb128: nope. I would put your branch into the same silo with the others [09:10] Saviq, also inside ScopeOverviewFavourite [09:10] mzanetti, that's not going to be a branch but a source upload, but those changes don't need to be synced do they? [09:10] Saviq, why are we not using a Row but we're calculating each x of the repeater? [09:11] mzanetti, you can land your change and then I can do an upload to drop the override later [09:11] tried with Row, works fine [09:11] seb128: no, they don't really need to. we just need to make sure my changes are first [09:11] right [09:11] seb128: ok. I'll ping you again about it when my stuff has landed [09:11] mzanetti, thanks [09:12] Cimi, most probably because of transitions [09:12] Cimi, but you'd have to talk to Albert [09:13] Saviq, ok [09:39] Saviq, tags [09:39] 0.1.16 loves u [09:40] Cimi, I can't do anything until it gets into trunk [09:40] ok [09:40] Saviq, oh, you mentioned tags yesterday, just remembering that ... what's the issue with those? [09:40] seb128, I pulled settings, there's like a hundred of ? tags [09:41] Saviq, could be, I don't usually look at those so I probably just never noticed [09:41] seb128, http://paste.ubuntu.com/8298371/ [09:41] Saviq, does it create any practical issue? [09:41] do you know where they are coming from? [09:41] seb128, they look citrain-y, but must've come from some other project [09:42] seb128, some bad merge or so [09:42] k [09:42] seb128, maybe the ubuntu branch? [09:48] Saviq: FYI. had a chat with Vesa. He thinks its not good to have gmail, facebook etc in there anyways and we should keep the current list of defaults. [09:48] mzanetti, well... what about gallery, camera? [09:48] they were there before [09:48] sure we shouldn't have gmail or facebook [09:48] Saviq: howdy. remember talking about the lag opening the indicator panel? [09:49] dednick, hey, welcome back, hit me [09:49] Saviq: thanks :) [09:49] mzanetti, anyway, if they're good with the current default set, fine by me [09:49] Saviq: it's because the models for the menus are synchornous. [09:49] dednick, yeah just read your reply [09:49] Saviq: but making them async creates the problem where we open the menus and nothing is there [09:49] for a second or so.. [09:50] Saviq: right. [09:50] dednick, sooo... why does it takes so long to sync? unitymenumodel slow? it's not like there's a huge amount of data [09:50] Saviq: i was wondering if it would create "that" much effect not disconnecting from the models [09:50] Saviq: no, it's actually not the models. the models run all the time [09:50] dednick, ah so just the delegates? [09:50] Saviq: it's actually the listviews adding the delegates [09:51] dednick, ouch, sounds excessive indeed [09:51] dednick, but well, should be easy to check what's the impact memory-wise [09:51] dednick, we sure need to set them invisible, but we might be able to keep them around inded [09:51] +e [09:52] Saviq: ya, we can probably just set the entire panel menu to invisible [09:57] Saviq: camera and gallery are there [09:58] http://paste.ubuntu.com/8298483/ [10:03] Saviq, can you pls repeat how to sbuild armhf? I tried sbuild -d utopic-amd64-armhf-shm but it build amd64 debs [10:04] Cimi, bug #1353855 [10:04] bug 1353855 in Mir "Explicit g++ 4.9 dependency breaks cross-building" [Medium,In progress] https://launchpad.net/bugs/1353855 [10:04] Cimi, and sbuild --help [10:06] Cimi, -d is *distribution* [10:06] Saviq, yes saw that [10:06] using -c [10:06] Cimi, you want -c for *chroot*, and you need to pass --host=armhf, otherwise it'll build amd64 [10:06] Cimi, how's it supposed to know what arch to build otherwise [10:06] ah damn [10:06] so what is the command? [10:07] -c utopic-amd64-armhf-shm --host=armhf [10:07] Saviq, wonderful, thanks [10:07] Cimi, but you need to drop the 4.9 dependency, otherwise it'll fail to install build deps [10:08] Saviq, where do I drop it? unity8? [10:08] Cimi, in whatever you're building [10:08] ok [10:08] Cimi, just read, please [10:09] Saviq, it doesn't explain which pkg I have to replace, so I just change -4.9 to 4.9 or just g++ ? [10:10] Cimi, drop it altogether [10:12] dednick, we don't want async then [10:12] https://code.launchpad.net/~nick-dedekind/unity8/indicator-polishing/+merge/228700 [10:12] dednick, all indicators are async, not only message [10:12] Cimi: it's too laggy when opening if we dont [10:12] dednick, well, we need to figure out a solution [10:12] Cimi: yeah, i'm trying to now. [10:15] Trevinho, do you know about a bug that apps scope doesn't notice newly installed apps? [10:31] Saviq, did the proposed branch have albert branch to cache more stuff? [10:31] Cimi, no, see silo [10:31] Saviq, ok [10:33] Saviq, ok so, the reason why the horizontal swipe is slower than the vertical one, is just that our finger moves quickly in a shorter space than the vertical swipe [10:33] maybe not entirely though, carousel is still so fast [10:34] actually carousel speed is tweaked further if I remember [10:34] it is [10:38] Holas [10:52] Saviq: so, the indicators when not loaded with messages or transfers takes up about 1-2MB. with a 3 completed transfers and 10 messages it was about an additional 5MB [10:53] dednick, right, so as long as we use ListViews there, we should be fine [10:53] so as not to create 100 messages in case someone has that many [10:54] dednick, sure it'd be better if we could create the delegates dynamically, but it'd have to be much faster [10:56] dednick_, what did you see last? [10:57] Saviq: so as not to create 100 messages in case someone has that many [10:57] dednick, sure it'd be better if we could create the delegates dynamically, but it'd have to be much faster [10:57] for now I think we're fine with not unloading them [10:57] Saviq: ya. perhaps the limitation should be on the backend [10:57] dednick_, backend doesn't know much about the UI though [10:58] Saviq: in fact there may already be one. [10:58] we're going that direction in the dash as well (keeping more in mem) [10:58] Saviq: meh. hard to limit on frontend since they're all the same. [10:58] i mean indicators are all the same, so hard to say max of x menu items [10:59] Saviq: but then this is all going to be changing again anyway right... :) [10:59] dednick_, exactly, and different delegates, different sizes etc. [10:59] :(!! [10:59] dednick_, what I meant is that a ListView with cacheBuffer: 2*height or something should allow us to keep memory usage low while improving the UX a lot [11:00] Saviq: backend limits age, so it should limit number as well [11:00] dednick_, yeah, but we can't rely on that, we need to make sure in UI that stuff works still [11:00] Saviq: right [11:00] dednick_, so let's go for that [11:00] Saviq: yeah, i was looking at that cachebuffer thing. [11:01] dednick_, items in cachebuffer have the important property of being created asynchronously [11:01] dednick_, so if you're not scrolling too fast, things should be smooth [11:02] dednick_, you might even change the cacheBuffer on indicator open [11:02] dednick_, so that we don't hold them in mem all the time, but create them async on indicator open, so that we have a buffer out of view [11:03] dednick_, one important thing though - the ListView is anchored to the bottom isn't it? kinda means stuff is destroyed as you close and recreated on open anyway [11:03] because as you drag the panel down, its height grows [11:03] but maybe that's fine, as long as we have some cacheBuffer and the experience is good [11:04] Saviq: ok, well i'll give it a go === MacSlow is now known as MacSlow|lunch [11:09] Saviq: mh, not that I recall... It used to work. Maybe sometimes it take some time... [11:10] Trevinho, I installed chrome when I pung you, still not there [11:11] Saviq: mh, ok killing the scope fixed it? [11:11] Trevinho, yes [11:12] /food [11:12] mh, wondering if some caching caused that... I'll check that, mh3 would have been more helpful here :/, but I'll look === Malsasa_ is now known as Malsasa [11:23] pete-woods, if you want, I got a script for syncing bug statuses between project and ubuntu package [11:23] and creating a Ubuntu task where applicable [11:23] Saviq: thanks. I just disabled bug reports on the upstream project like was mentioned in that bug :) [11:24] apparently that's our policy now [11:24] pete-woods, yeah, but you still have bugs that are only against libusermetrics [11:24] oh, I thought I moved them [11:24] pete-woods, ah well, maybe you have enough bugs that you did it manually is all :D [11:24] there were only about 10 bugs [11:24] pete-woods, yeah ;) [11:25] pete-woods, so yeah, you're good [11:25] (no-one knows the project exists, ssssh ;) ) [11:25] * Saviq still was unable to un-affect all the unity8 bugs, apparently not possible via the API :| [11:25] that's a shame [11:25] I'd kinda expect LP to have a bulk-edit feature in the UI [11:25] like other tools I've used in the past [11:27] they do, the API ;P [11:34] dandrader, try time make -B -j15 and time make -B -j8, the latter is faster to me [11:38] Saviq: it's pretty jerky with the cacheBuffer lowered [11:42] Cimi, "real" and "user" are faster but "sys" is slower. Whatever that means. But yeah, seems to be a bit faster [11:50] Cimi, got another small one, if you have some spare minutes: https://code.launchpad.net/~dandrader/unity8/buildWithoutWarnings/+merge/233736 [11:51] dandrader, I have spare time [11:51] currently just testing flickable speeds [11:51] "just" [11:51] is experimenting === MacSlow|lunch is now known as MacSlow [12:00] dednick_, in what case? [12:00] Saviq: scolling messages. esspecially if you drag over bounds. it gets a bit screwed up. i think the item heights are changing. [12:01] dednick_, that's weird === seb128_ is now known as seb128 [12:20] Saviq: not really when you think about it. http://paste.ubuntu.com/8299420/ [12:20] i meant that the items height's were changing on purpose. [12:20] dednick_, just make the Loader height: 40 [12:21] Saviq: Component.onCompleted: height = 60 [12:21] Saviq: need the loader height to follow the item height [12:21] dednick_, right [12:22] dednick_, yeah, ListView and changing heights isn't too great indeed [12:22] Saviq: ya. the height should probably expand up if the item is above :/ [12:23] bit of a pain [12:33] Saviq: even weirder that it doesn't happen if you use boundsBehavior: Flickable.StopAtBounds [12:33] dednick_, not really [12:33] dednick_, well, when you overshoot [12:34] Saviq: i guess when you flick over it's removing more items then readding when it comes back into view [12:34] dednick_, things get recreated [12:34] dednick_, yeah, that's really a bug in ListView, it should never take overshooting into account [12:34] because that's just wasting CPU [12:34] and some other things === _salem is now known as salem_ [12:53] Saviq: i raised a bug with qt about that issue. I think i'll just turn off the flickable overshoot until that gets fixed and mark it as fixme ? [12:54] dednick_, if that's the only problem, yeah [12:54] Saviq: meh. it's still a bit jerky just scrolling around [12:55] dednick_, well, but it's rarely going to be the case anyway, when people have dozens of messages? [12:55] or transfers? [12:55] Saviq: ya [12:56] sometimes restarting unity8 I have this error terminate called after throwing an instance of 'boost::exception_detail::clone_impl >' what(): Exiting Mir! Reason: Nested Mir and Host Mir cannot use the same socket file to accept connections! [12:57] Cimi, environment got screwed [12:57] Cimi, it's trying to connect to and listen on the same socket [12:58] Saviq, yes indeed [12:58] Saviq, I was having the same issue when the wizard had issues with the upstart job [12:58] Saviq, but all I am doing is restart unity8 as phablet [12:58] and not the first time happens [12:59] Cimi, yeah, we need to look at our upstart job, something wonky must be happening on restart [13:01] btw I replaced the flickable with both repeater and listview insode scope favorites, still performance is not comparable to cardHorizontalList [13:01] I don't know what exactly is happening here, but is not right [13:11] dandrader, mzanetti found a reliable way to reproduce the mouse input issue (in the dash, not in shell) [13:12] dandrader, I imagine your qtmir fix fixed it for the shell, but then the apps can still get into that state can they not? [13:12] dandrader: hi. yeah, the one I told you last week. I also wrote it to the bug report now [13:12] dandrader, https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1295623/comments/20 [13:12] Ubuntu bug 1295623 in unity8 (Ubuntu) "Sometimes input breaks and only edges are responsive" [Critical,In progress] [13:12] Saviq, likely. racarr is also working on the mir side of the problem [13:14] if I'm not mistaken, he found out that android-input, or the evedev input device, was ending touches in a weird way. At least on krillin [13:14] dednick_, so... on another topic, we have bug #1365530 again [13:14] bug 1365530 in unity8 (Ubuntu) "Time in indicator out of sync on resume from suspend" [High,Triaged] https://launchpad.net/bugs/1365530 [13:14] dandrader, yeah, prolly related to the buttons [13:15] Saviq: happy days [13:15] Saviq: or is that where it doesnt update for a little bit? [13:15] dednick_, no, it will be many minutes behind until the next update comes [13:16] dednick_, yeah, I think another symptom is that the blue LED lights up only after you turn the screen on (I expect the LED thingy in unity8 doesn't get the icon update) [13:16] meh [13:17] stupid model cache [13:31] mterry: hey, would you know of a nice way for me to bring up a device replacing USC with a custom mir server? Is usc hardcoded into lightdm? [13:33] greyback_, there's a usc-wrapper for our session [13:33] greyback_, check out dpkg -L ubuntu-touch-session [13:34] Saviq: ah there it is. Thanks! [13:35] greyback_, you can change the command lightdm uses, hold on [13:35] greyback_, look at how /usr/share/lightdm/lightdm.conf.d/52-ubuntu-touch.conf does it [13:35] /usr/share/ubuntu-touch-session/usc-wrapper <- that's it I guess [13:35] greyback_, sure you can edit that directly [13:35] Saviq, are we moving towards just bugs in the package? [13:35] yep that file calls the wrapper [13:35] greyback_, didn't know how hacky you wanted :) [13:36] Cimi, we're not moving, we're there [13:36] mterry: just for my testing, nothing more. thanks :) [13:36] Saviq, just seeing you removing unity8 from bugs [13:37] Cimi, yeah, project bugs were turned off last week already, I just didn't know how to batch-remove the tasks === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader [14:38] so there seems to be two different issues for our weird flicking [14:39] 1st there is input lag between the touch event and the activity on screen [14:40] this is probably mir related [14:40] 2nd qt flick detection seems to be bugged, swipes are not consistent, sometimes goes fast sometimes everything stops or moves slowly [14:41] this could be both a problem in qt and/or mir not sending all touch events (or delaying them) [14:41] greyback_, dandrader Saviq ^ [14:42] Cimi, there's a branch for the input lag [14:42] Cimi, [14:42] great, I can try that to see what changes for issue 2 [14:42] https://code.launchpad.net/~vanvugt/mir/butter/+merge/230767 [14:43] Cimi, it should be in a ppa I think [14:43] Saviq, mir devel? [14:43] Cimi, https://launchpad.net/~mir-team/+archive/ubuntu/staging [14:44] it's missing qtmir though, and has unity-mir which could be dropped [14:44] camako, is ↑↑ the PPA for Mir devel? [14:45] Saviq, yes [14:45] Saviq, I am pretty sure problem 2 has still some origins inside qt as well, because while I don't have 1st in desktop under X, I have sometimes a weird flick on desktop too === salem_ is now known as _salem [14:45] Cimi, not impossible [14:46] Saviq, and explains the recent patches we saw regarding my qt bug [14:46] camako, you can probably drop qtmir from there, but you should add qtmir/devel-mir-next there probably [14:46] Cimi, the patches only dealt with the initial threshold AFAICT [14:46] Saviq, yea we haven't fixed it after the unity-mir qtmir switch [14:47] eerm I meant drop unity-mir [14:47] Saviq, indeed, with small/fast swipes the visual flick seems badly calculated [14:47] Saviq, if you do a long swipe, the flick appears natural [14:48] Saviq, if you start swiping like crazy, and fast, the flick gets stupid [14:48] Cimi, yeah, but I don't think the patches would help, they're just about about the initial threshold before a flick even happens [14:48] Saviq, on android I can flick like crazy and it does not mess up the current flick [14:49] Saviq, here it feels like some detections are wrong [14:49] dednick_, on your indicators polishing, is it possible to keep the models always updated, and just hide the UI with loaders? [14:49] dednick_, what is the reason for the delay? [14:50] Cimi: yes, well that's the way we decided earlier [14:50] Cimi: because they take memory [14:51] dednick_, but seeing the indicator message green, opening the menu and finding it empty is even worse [14:51] hm i thought butter already landed? [14:51] dednick_, maybe we can add some condition to the loading, loading when there is new content? [14:52] dednick_, also, what take up memory, the indicators model or the UI? [14:53] anpok, it did... in 0.7.0 [14:57] Cimi, ↑ ok so nothing new in Mir devel, the input delay should be better with Mir 0.7.0 already [14:58] Cimi, but it's a hw issue to some extent [14:59] Saviq, I know, but not that bad [14:59] Cimi, you'd be surprised ;) [15:00] Cimi, some time ago we installed a multi-touch painting app on manta just to check what to expect, it was really bad [15:00] on android, that is [15:01] Saviq, https://code.launchpad.net/~cimi/+junk/multitouchtest [15:01] hah [15:02] Saviq: am sure you're busy, but did you ever get change to take a peek at lp:~gerboland/unity8/dbus-async just to see if you approve of where I put the shared library? [15:02] old stuff [15:02] greyback_, I even replied :) [15:02]