=== howefield is now known as howefield_afk === jamesh_ is now known as jamesh === chriadam|away is now known as chriadam [01:31] is there any event for exiting the application? [01:33] I wan to save some data when my app exits or terminated. is there any way to capture such an event? [01:33] You can hook into lifecycle events. [01:34] Bouncy bounce! [01:36] liuxg_: If you're actually here this time, lifecycle events are what you're interested in. [01:36] RAOF, what are the lifecycle events? could you please highlight? [01:37] liuxg_: But you should really be continuously saving state unless it's super-expensive. [01:38] RAOF, I am trying to capture "Component.onDestruction", which works on the desktop, but it does not work on the phone when the app is closed. [01:39] Yeah, you won't get that on the phone. [01:39] Because you're first SIGSTOPped and then may at some point later be killed. [01:40] RAOF, normally an app is pushed the background, and it is suspended. What about the case when the app is in the foreground? [01:41] * RAOF can never find what he's looking for on the SDK site :( [01:41] liuxg_: You can't close a foreground app on the phone. [01:42] Well, the app can, but there's no standard chrome for it. [01:42] The only UI for closing stuff is going to the switcher and then swiping the app away; once you're at the switcher, the app is no longer in the foreground. [01:44] I know what you want at the Mir level, but urgh our documentation is endlessly frustrating. [01:44] Sorry. === chihchun_afk is now known as chihchun [07:18] good morning [08:15] good morning! [08:42] cking: Ping [08:42] sturmflut2, hiya [08:49] Good morning all; happy Cars Day! 😃 [09:08] hello again [09:13] how the bug reporting using `ubuntu-bug' should work on the phone? [09:14] I tried `adb shell ubuntu-bug unity-scope-click' and after a while I was asked what I'd like to do [09:14] Sleep_Walker: https://wiki.ubuntu.com/Avengers [09:14] you can't use ubuntu-bug on click packages [09:15] and neither `Send report' nor `View report' does anything [09:15] only on debian packages. [09:16] unity-scope-click looks like debian package [09:16] ok, super [09:16] you should probably adb shell, _then_ run ubuntu-bug unity-scope-click. [09:17] OK, I'm doing that now (+ bash -x to see where it hangs) [09:17] (although it works for me) [09:17] aha! [09:17] yeah [09:17] http://paste.ubuntu.com/11667905/ [09:17] less is misbehaving when invoked directly [09:18] please, fix your Avengers page [09:18] adb shell ubuntu-bug unity8 [09:18] ok, will do :) [09:18] thanks :) [09:18] done [09:18] np [09:18] appreciate the feedback [09:19] 👍 === chriadam is now known as chriadam|away [09:20] nice, duplicate [09:22] btw. can I somehow disable security for ADB? It's somehow annoying... [09:23] the fact that you have to unlock the phone? [09:23] use ssh :) [09:23] yes [09:23] Sleep_Walker: "ad shell" has a lot of issues, you can use "phablet-shell" instead. [09:23] Sleep_Walker: http://sturmflut.github.io/ubuntu/touch/2015/05/08/hacking-ubuntu-touch-part-5-adb-shell-vs-phablet-shell/ [09:24] he has a blog post for all occasions [09:24] ssh is good option, I don't have phablet-shelll [09:24] :D [09:24] *phablet-shell [09:24] popey: It's the 5 P's [09:24] popey: "Proper preparation prevents poor performance" [09:25] sturmflut2: that's not the 5p's I learnt [09:25] davmor2: Now you've got my attention [09:26] sturmflut2: "Preparation prevents P*** poor performance" [09:27] you have it short, don't you? ;) [09:27] davmor2: Well, 4 of your 5 P's are identical to mine, that's about 80%, and 80% is about the same as 100%, so yours is identical to mine [09:27] Science! [10:11] barry: hey, do you know if system-image 3.0 got into any other channel than devel-proposed? [10:35] nerochiaro, I commented on https://code.launchpad.net/~uriboni/webbrowser-app/keyboard-navigation/+merge/260183 [10:40] oSoMoN: when you say "recentView" you mean the history ? [10:44] nerochiaro, no, I mean the tabs list view (the id in Browser.qml is "recentView") [10:45] I know the id is kinda confusing, that’s because long term the history view will be part of the same view [10:45] oSoMoN: how do you get there ? [10:45] oSoMoN: it is kinda confusing [10:46] nerochiaro, recentView is supposed to be an aggregated view that has the list of open tabs and the history view, all in one [10:46] that’s the original design for the bottom edge, but it hasn’t been fully implemented yet [10:47] oSoMoN: ah, i tried bottom edge swipe but nothing comes up [10:47] nerochiaro, bottom edge swipe is on touch devices only, on desktop you access it from the drawer menu [10:50] oSoMoN: i feel very stupid but on desktop in the drawer i have history, open tabs, new tab, settings and private mode. which one is the recent ? [10:51] nerochiaro, don’t feel stupid :) "open tabs" is the one you’re looking for [10:52] oSoMoN: ah, very aptly named [10:52] (as I said, "recentView" currently corresponds to the list of tabs) [10:52] allright, so totally ok to get out of there and out of settings with ESC [10:56] oSoMoN: and regarding the way we navigate across tabs, we had discussed this already i think. the tabs model does not keep a list and you can simply set the index of the current one. instead when you set the current one to any index, it gets moved to the top of the list. [10:56] nerochiaro, yes [10:56] oSoMoN: so there is no way to properly navigate in the list, other than "forward" [10:57] oSoMoN: by taking the last tab on the list and making it current [10:57] nerochiaro, are you referring to my comment about the implementation of Ctrl+Tab? [10:57] oSoMoN: yes [10:57] ok, got it [10:59] nerochiaro, I guess that makes sense then, but could you maybe add a comment to explain that in more details? The "navigate to next tab" comment is not really accurate [10:59] oSoMoN: sure thing [11:24] what kernel is running on ubuntufor devices === MacSlow is now known as MacSlow|lunch [11:25] srix, the source comes from android (has to) but it is patched for security (adding apparmor for app confinement etc) and has a good bunch of fixes from canonical [11:25] (well, the latter is true for all supported phones :) ) [11:50] Does anybody know why forkstat shows so much activity even when the phone is locked? -> http://paste.ubuntu.com/11670349/ [11:53] Looks like lightdm tries to start unity-system-compositor in a loop? But why? [11:58] this does not look right === alan_g is now known as alan_g|lunch [12:01] anpok_: Yeah, right? [12:02] sturmflut2: how can I add apps to your wishlist? Fish Fillets (Game), ScummVM (must have adventure game interpreter), GPS status application with detailed info), Simple GPS logger, simple music player with filesystem hierarchy view, StarDict/GoldenDict offline dictionary, Locker (or better to have this in top menu), Synergy or at least VNC, OSM offline viewer? [12:03] sturmflut2: i dont know why this hapens.. you could file a bug with ligtdm/u-s-c or dig deeper... [12:03] *Locker - prevent phone from sleep [12:04] Sleep_Walker: The easiest way is to just tell me, but you can also go the "official" way and create an issue at https://github.com/Sturmflut/sturmflut.github.io/issues [12:05] sturmflut2: are you sure that this isnt just process activity and not fork sys calls? [12:07] anpok_: Let's ask cking, he wrote the tool, but it should be actual calls to fork() and exec() [12:07] hm just reading a documentation about it.. [12:07] so yes looks like syscalls.. [12:08] btw it does not happen for me on mako [12:08] but on screen on it happens [12:08] anpok_: Keeps happening on my bq after I've restarted the phone, on OTA-3.5 [12:09] and it sometimes happen when i interact with the device .. [12:09] sturmflut2, it may be those forks are just handler threads being spawned off - they have a short life span [12:10] some lower level library above libc doing that [12:11] cking: Ah, right, it's just a fork() and no exec(), correct? [12:11] hm ok still interesting.. could be libdbus.. [12:11] i guess it's a thread in a support library that libdbus is using, possibly some callback with a timeout [12:12] sturmflut2: does that happen all the time? [12:12] could you also monitor dbus.. [12:14] perhaps attaching health-check or strace to it will prove to be instructive [12:14] cking: hm why does lightdm show up in the log? [12:14] anpok_: Sometimes it stops for a short amount of time, but in general it happens all the time, yes. And it happens a lot, like multiple times per second. [12:14] anpok_, because a thread is being created that does something and then the thread exits [12:15] cking: so the thread is created inside lightdm?? [12:15] That's what confuses me as well [12:15] anpok_, yep, but probably one of the libraries it uses [12:15] it may be just a timer [12:15] implemented in a crude way [12:15] ah ok so it is rather something inside lightdm [12:16] i was mentally checking through what usc does wrt thread starting/stopping === howefield_afk is now known as howefield [12:21] anpok_, yep, i bet it's a lower level library it is using, normally is something like that [12:22] proabably libglib, it normally is that [12:23] Okay, the lightdm process that shows up as a parent on my phone is currently PID 1763. If I "strace -ttt -y -f -p 1763", I see nothing. [12:24] sturmflut2: if it really tries to start usc again. usc will exit because the mir-socket is still there.. [12:25] anpok_: Yep, I just dropped the command line that shows up into phablet-shell, see http://paste.ubuntu.com/11670740/ [12:25] anpok_: Also note the "Warning: ignoring unrecognised arguments: --vt 1" [12:30] you could try health-check on it, it's like strace on steroids === _salem is now known as salem_ === pat__ is now known as pmcgowan [12:40] cking, anpok_ : I have the feeling that forkstat doesn't tell me what I think it does. "health-check -p lightdm" doesn't show any calls to fork() or clone(), and when I set a filter to "sys_fork", "sys_clone" and "sys_exit" with ftrace, lightdm doesn't even show up in the output. According to ftrace the only processes at "fault" are unity-system-compositor, unity8-dash and unity8 for using clone() to create short-lived [12:40] threads. [12:40] (at least that's what I can tell with my capabilities) [12:43] sturmflut2, what device are you using? [12:43] cking: bq Aquaris E4.5 [12:43] ok, lemme have a peek once I reflash it === chihchun is now known as chihchun_afk [12:44] forkstat only reports what the kernel sees, not sure how that gets picked up when procesesses are very short lived [12:46] bah, phone is flat, gimme 30 mins while I pump it with charge === tvoss is now known as tvoss|test === tvoss|test is now known as tvoss === marcusto_ is now known as marcustomlinson [12:49] cking: I think the "fork $PID child" and "exit $PID" messages are correct, but the "fork $PID parent" ones don't make sense. Why does parent "lightdm" create a child called "unity-system-compositor"? Without a call to exec()? [12:50] yeah, that sounds wrong [12:50] The example on http://kernel.ubuntu.com/~cking/forkstat/ looks much more consistent in that regard [12:51] Let's dig deeper... [12:51] sturmflut2: that --vt 1 thing does not hurt.. [12:52] 09:42:49 fork 19268 parent /bin/sh /usr/bin/lesspipe [12:52] 09:42:49 fork 19269 child /bin/sh /usr/bin/lesspipe [12:53] cking: That's what I would expect [12:53] this means "proceess 19268 is the parent, it is /bin/sh /usr/bin/lesspipe", and "process 19269 is the child, child is called /bin/sh /usr/bin/lesspipe (for the moment)" [12:54] cking: Yep, that's what I would expect, both should have the same name after the fork() [12:54] can you point out the exact two lines that are the issue === alan_g|lunch is now known as alan_g [12:55] 13:49:18 fork 1785 parent lightdm [12:55] 13:49:18 fork 22383 child unity-system-compositor --disable-overlays=false --spinner=/usr/bin/unity-system-compositor-spinner --file /run/mir_socket --from-dm-fd 9 --to-dm-fd 13 --vt 1 [12:57] ok, i'll git into that in a moment [12:58] sturmflut2, OK, I can reproduce this, I'll debug it now [12:59] cking: \o/ [12:59] sturmflut2: 12:58:55 fork 1468 parent lightdm [12:59] 12:58:55 fork 4482 child unity-system-compositor --disable-overlays=false --spinner=/usr/bin/unity-system-compositor-spinner --file /run/mir_socket --from-dm-fd 10 --to-dm-fd 13 --vt 1 [12:59] 12:58:55 exit 4482 0 0.001 unity-system-compositor --disable-overlays=false --spinner=/usr/bin/unity-system-compositor-spinner --file /run/mir_socket --from-dm-fd 10 --to-dm-fd 13 --vt 1 === tvoss is now known as tvoss|test [12:59] ps -ef | grep 1468 [12:59] root 1468 1 0 12:57 ? 00:00:00 lightdm [12:59] root 1496 1468 1 12:57 ? 00:00:01 unity-system-compositor --disable-overlays=false --spinner=/usr/bin/unity-system-compositor-spinner --file /run/mir_socket --from-dm-fd 10 --to-dm-fd 13 --vt 1 [12:59] root 1575 1468 0 12:57 ? 00:00:00 lightdm --session-child 10 16 [12:59] root 4831 3752 0 12:59 pts/43 00:00:00 grep --color=auto 1468 === tvoss|test is now known as tvoss [13:01] sturmflut2, so pid 1496 has a parent 1468, which is lightdm according to ps, and that's the kind of fork/exec activity I'm seeing with forkstat [13:04] cking: Let me think for a moment [13:10] Argh, I think I know what's happening [13:14] cking: the Process Events Connector plugs into do_fork() in the kernel, and do_fork() is called by *both* the fork() and the clone() syscalls. But they apparently never added a PROC_EVENT_CLONE message, instead a call to clone() generates a PROC_EVENT_FORK message [13:16] jgdx, do you/Ken have plans for u-s-s in vivid-overlay? [13:16] cking: if you do an actual fork(), the parent_pid is that of the calling process, the child_pid is a newly allocated one, and the child is a copy of the parent, so the cmdline is the same [13:16] jgdx, landing wise I mean [13:16] jgdx, I would like to backport a few wily fixes now that vivid is open again, don't want to conflict with your apn work or whatever is in silos though [13:16] not sure if those are for testing or landing [13:16] cking: But if you do a clone(), the parent_pid is that of the process which initially forked() the calling process [13:17] sturmflut2, oh, that expains it, urgh [13:18] cking: So lightdm actually called fork() and exec() to create unity-system-compositor as a child in the very beginning, but unity-system-compositor later goes and calles clone() to create a thread, and at that point everything goes wrong === MacSlow|lunch is now known as MacSlow [13:19] which means forkstat is a bit fatally flawed on the clones, urgh [13:19] seb128, funny you should mention it, I'm trying to coordinate this here https://docs.google.com/spreadsheets/d/1F6eKzwpobzoAW1JyvFt_nacFltQiIMCniA793F1z2Lc/edit?usp=sharing [13:19] cking: Yep, and it also explains why I didn't see anything with strace and why ftrace painted a completely different picture [13:20] ok, well that's great you spotted that, I completely overlooked that :-/ [13:20] jgdx, oh, there are more I want to backport ... let me set up a few mr and add them to the table then ;-) [13:20] seb128, I just started. :) Perfect! [13:21] Tassadar: it hasn't yet. i'm waiting for someone to actually request it in another channel [13:21] cking: I'm actually happy that I apparently know enough about Linux to have spotted this. Now maybe we can find a solution. [13:21] barry: good, thanks [13:21] hope I didn't break the other channels with my updates actually, heh [13:21] probably not though [13:26] cking: I think the error is here, https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/kernel/fork.c#n1595 , where it calls proc_fork_connector() without any distinction between clone() and fork(). And then https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/connector/cn_proc.c#n67 goes and builds a PROC_EVENT_FORK message in both cases [13:28] sturmflut2, yep, that's what I was just looking at - however, I'm debugging a firmwar eblob issue at the moment so I'm finding it hard to follow mutltiple debug threads at the moment [13:29] i wonder if the semantics have changed or it's always been broken [13:30] cking: no problem, let's defer the discussion. I still have to write about your other tools ;) [13:30] sturmflut2, I'll have a think about the issue and see if I can figure out a workaround [13:31] unfortunately, this kind of stuff with short lived processes is inherently racy when getting info from /proc === howefield is now known as howefield_afk === chihchun_afk is now known as chihchun [13:47] seb128, ~ubuntu-branches/ubuntu/vivid/ubuntu-system-settings/vivid is what we target, right? [13:48] jgdx, no [13:48] jgdx, lp:ubuntu-system-settings/15.04 [13:48] jgdx, https://code.launchpad.net/~system-settings-touch/ubuntu-system-settings/15.04 [13:48] okay [13:51] seb128, why is that changelog so different? [13:52] jgdx, which one? [13:52] for the branch you just linked to [13:52] different to what? [13:52] jgdx, http://bazaar.launchpad.net/~system-settings-touch/ubuntu-system-settings/15.04/view/head:/debian/changelog [13:53] seb128, from that. Thanks [13:53] jgdx, vs http://bazaar.launchpad.net/~system-settings-touch/ubuntu-system-settings/trunk/view/head:/debian/changelog [13:53] jgdx, I'm unsure to understand the question :-) [13:53] seb128, no debian changelog here http://bazaar.launchpad.net/~system-settings-touch/ubuntu-system-settings/15.04/changes [13:53] was just wondering why [13:54] jgdx, ? http://bazaar.launchpad.net/~system-settings-touch/ubuntu-system-settings/15.04/revision/1416 [13:54] vs e.g. https://code.launchpad.net/~ubuntu-branches/ubuntu/wily/ubuntu-system-settings/wily [13:54] oh, changelog is autogenerated by the ci bot [13:54] not manually commited [14:01] Aquaris E5 lists "Video formats: .avi, .mkv, .mov, .mp4". Good to know what containers i can use...but what VIDEO formats does it support now [14:02] utack, sturmflut2 has made a list of supported codecs recently [14:02] do you know where? [14:02] utack: http://sturmflut.github.io/ubuntu/bq/2015/05/31/hacking-the-bq-part-3-supported-media-plugins-and-codecs/ [14:02] utack: at the bottom [14:02] there :D [14:02] thx [14:03] I KNEW this would come in handy at some point [14:03] haha [14:03] good thing you are doing aquaris job there [14:07] sturmflut2 do you know if it has hardware support for some codecs, like h264? [14:07] webm and vp* are probablly decoded in software? [14:09] utack: The current OTA-3.5 software release for the Aquaris E4.5 decodes H.264/AVC, DivX, H.263, MPEG-4, Sorenson H.263.1 (s263) and Xvid in hardware, through the Android drivers. All other video codecs, and *all* audio codecs, are currently decoded in software. On the E5 it's most likely the same [14:11] utack: Actually the MediaTek hardware can do VC1 and VP9 in hardware, and the Android drivers are there, but the Ubuntu side is missing as far as I know [14:11] * sturmflut2 looks at jhodapp [14:11] not bad, didn't expect that from mediatek [14:11] don't think qualcomm gives a crap about vp9 and vc1 [14:12] seb128, could we do one silo for all of those in the spreadsheet? [14:12] do you have anything to add there? [14:12] jgdx, I'm mp-ing vivid changes atm, not sure if you got the emails [14:13] jgdx, but yeah, that was my idea, flush the backlog in a "backport fixes" landing [14:13] sturmflut2 - i notice you were doing some baytrail linux stuff, did you ever get the wifi working, or know anyone that did? [14:13] utack: At least some of the Qualcomm chips do VC-1 and WMV-9 in hardware [14:13] ok [14:13] seb128, k [14:14] well tbh h264 is perfectly fine, since the soc vendor already paid the fees and x264 is a really good encoder, but longterm it would be great to shift to free codecs [14:15] jgdx, https://code.launchpad.net/~system-settings-touch/ubuntu-system-settings/15.04/+activereviews [14:16] MiW: I never got my tablet far enough to actually care about the WiFi, usually it crashes or freezes because of other issues. Sébastien Bourdeauducq (sb0) over in #m-labs apparently got WiFi working with the right firmware file for the Broadcom chip [14:17] seb128, done [14:17] brillant thanks for the tip [14:17] jgdx, thanks [14:18] sturmflut2 cking: re, what was the result on the forks? Was that a wrong output of the tool or does lightdm really attempt to restart u-s-c frequently or (less frequently depending on the lucky device)? [14:18] seb128, okay, I'll start mping as well [14:18] anpok_: The kernel sends the "wrong" netlink messages to forkstat [14:19] anpok_, looks like a feature in the way sys_clone is wrong ^ [14:19] ok [14:19] but we know that lightdm on sturmflut2s bq shows awkward activity for some reason~ [14:21] jgdx, which ones? [14:22] jgdx, I'm doing the locale compare and dash security [14:24] anpok_: No, that was a "false positive", lightdm doesn't do anything wrong. [14:24] oh [14:25] anpok_: unity-system-compositor, unity8-dash and unity8 create a couple of short-lived threads per second, that's all [14:25] i was curious because I never saw that when the screen is off [14:27] anpok_: To be exact, those three processes spawn threads exactly every 500 milliseconds each [14:29] Hi, do you know the release date of desktop convergence, and if it will work on a Nexus 4? [14:31] Z3: the first device with Convergence will be released by bq somewhere towards the end of the year, some say in October. [14:31] yes, no. [14:31] jgdx, k, I'm done with the backports I wanted to do, I'm unsure about the other ones from Ken [14:31] well, yes, maybe [14:32] sturmflut2 ok. And a las question: what's the release date of a Ubuntu Touch tablet with convergence? [14:33] jgdx, I'm happy to put that stack in a silo if you think we are good with those [14:33] Z3: I don't think we have any info on that. Maybe the announced bq device will be a tablet, who knows [14:33] sturmflut2 ok, thank you very much :-) [14:36] seb128, I got those from the backlog. Not sure they are all of the backports [14:36] jgdx, https://code.launchpad.net/~system-settings-touch/ubuntu-system-settings/15.04/+activereviews has the stack [14:40] Sleep_Walker: I added all your app suggestions to the next iteration of my list, which will be published in about three weeks. The dictionary app already exists I think, https://uappexplorer.com/app/knowndict.benyfu [14:41] thanks for tip! [14:41] Sleep_Walker: and popey is going to have ScummVM up and running in a couple of days :) [14:42] mcphail: What?!? [14:42] sturmflut2: do you actually have a schedule on when you update the posts? [14:42] hahah [14:42] popey: any progress? :) [14:42] nik90: I do now, the target is to publish at the end of every month [14:43] mcphail: not touched it since we last looked at it [14:43] may 22nd according to my log [14:43] popey: just wanted to check. Have my finger poised over the buy button on gog.com [14:43] hah :) [14:44] hum [14:44] I think I bricked my phone [14:44] bq aquaris 4.5 ubuntu edition [14:44] hold down the power button for longer than you think you need to [14:44] almost always fixes it :) [14:44] 10s plus [14:44] it reboots alright, only I can't reach recovery mode anymore [14:45] it boots into this after the operator logo: http://www.androidguys.com/wp-content/uploads/2014/04/Samsung-Grand-hard-reset.jpg [14:45] tried to flash it with android [14:45] oh [14:45] cant help there [14:45] Bargh, I wanted to also do a "My favourite new apps" article every month, but there are now so many updates to the app store that I have a hard time keeping up [14:45] I can't believe a phone from 2015 can be brickable [14:46] lol [14:46] Silex: m( [14:46] still goin on about that [14:46] Silex: not convinced it's completely bricked, but we don't really support android here ㋛ [14:46] try #ubuntu-touch [14:46] ROFL [14:47] this is just goin around in circles [14:47] screwsss: what are you after? [14:47] the latest build of ubuntu [14:47] sturmflut2: too many apps is surely a good thing? [14:47] screwsss: for what device? [14:47] windows [14:47] popey: You can actually *really* brick it if you try to flash Android on it, just kill the Preloader and that's it [14:47] screwsss: troll better [14:48] sturmflut2: okay, so it is brickable. Probably what I did then [14:48] do i need to be a linux expert to be able to comfortably use unbuntu [14:48] it's a shame because there *is* some kind of bootloader, only that /dev/ttyACM0 isn't talkable long enough when I plug the phone [14:48] is there an efs partition on those phones similiar to samsung phones? [14:49] wow, that animation definitely doesnt exist on the phone by default [14:49] where did you get the recovery from ? [14:49] ogra_: the BQ Aquaris 4.5 android edition [14:50] it's the same hardware [14:50] we dont ship with android though [14:50] yes, but different partitionin and some minor HW diffs (screen etc) [14:50] figured it would be easy to "just put android on it" [14:50] ooooh [14:50] then I'm fucked indeed [14:50] you need to use the flash tool that bq provides [14:50] contact bq, they may be able to help [14:50] the boot partitions including recovery are the same [14:51] just the system partitions are differently sized [14:51] ogra_: the windows tool? alright yeah let's give it a try [14:51] I just need to install a windows box :( [14:51] their flash tool actually flashes on a low level and re-writes the partitioning [14:51] ogra_: shame we can't do the same on linux [14:51] and virtualbox USB support isn't there yet [14:51] Silex: The SP Flash Tool is available for Linux somewhere [14:51] Anyway, that gives me hope. Thanks! [14:52] sturmflut2: it keeps getting better :) [14:52] * Silex googles [14:52] bq should put a big warning label on those downloads [14:53] i dont think there is actually a way to brick the phone if you dont do overly insane things like dd'ing /dev/zero into the disk [14:53] you should always get to some fastboot mode [14:54] ogra_: "Some people on the internet" apparently actually did it, they flashed the wrong stuff with the SP Flash Tool and then the preloader was corrupt. [14:54] ogra_: that's what I thought too but I booted into fastmode alot of times before flashing, and now it just doenst' work. Only reboot does. I'll try flashtool.net [14:55] sturmflut2, right, thats about the same level as dd'ing /dev/zero into your MMC [14:57] ogra_: But the SP Flash Tool has a fancy GUI and can be downloaded from the bq website. Actually bq offers it as the "default" way to do a hard reset, so people probably get the impression that it's safe in all cases [14:57] yeah, it sadly has some risky buttons [14:59] what is mir in laymens terms [15:01] i thought re-paritioning a phones main memory could really fuck it [15:01] well, the images have fixed partition tables that are actually created for that specific device [15:02] so Silex is in the clear? [15:02] ogra_: At the very end the phone is unbrickable, right, even if you zero the eMMC flash. MediaTek SoCs have a small internal ROM with a minimal bootloader that is executed before it boots from flash and initializes the hardware, and if you open the phone there apparently are some JTAG connections on the circuit board which can be used to flash the eMMC. [15:02] ogra_: But that's most likely a case for customer support then [15:02] indeed, but that requires some more HW effort :) [15:03] screwsss: I have more hopes than before but I think there's still little chance of really recovering it [15:04] try it [15:04] yeah, when at home I definitly will [15:04] just do it. dont let your dreams just be dreams [15:04] obv you have nothing to lose [15:04] so whats this mir business [15:05] screwsss: It's basically the thing between your applications and the graphics drivers [15:06] sturmflut2, ogra_, popey: thanks! [15:06] np [15:06] good luck ! [15:08] cking: Is power-calibrate usable on the E4.5? [15:09] sturmflut2, I'm not 100% sure, I'd avoid that one if possible [15:09] it's Work-In-Progress [15:11] cking: Okay, it would have been a major change to my setup since it needs the device to discharge the battery, and all my communication goes over USB, which automatically charges the battery [15:12] sturmflut2, I've not 100% ensured it produces reliable answers, so it's not production quality yet [15:12] cking: I guess the same applies to powerstat? [15:12] sturmflut2, powerstat is OK-ish, as long as the battery is linear-ish, which is may not be [15:13] and it's only useful on bigger devices, like laptops [15:13] *note* [15:14] cking: I really appreciate the effort you put into your man pages BTW [15:14] sturmflut2, thanks, I try to make them useful, rather than "minimal" === chihchun is now known as chihchun_afk [15:33] * sturmflut2 already loves smemstat [15:40] sturmflut2: that did not explain it well for me but neway [15:40] anyways [15:41] screwsss, it is a display server ... it also has its own IRC channel in #ubuntu-mir [15:43] screwsss: Your applications want to draw something on the screen, and the graphics drivers know how to draw things on the screen, but you can't allow every application to directly access the graphics drivers. That would end in chaos. You also want additional features like window management, you want to switch between applications and so on. So you need a component between the applications and the graphics drivers that does [15:43] all this advanced staff. [15:43] screwsss: https://en.wikipedia.org/wiki/Display_server [15:43] screwsss: The Wikipedia page even lists Mir [15:43] s/staff/stuff/ [15:44] ah i see [15:46] (From the live Q&A) Okay, no news on an official Ubuntu tablet :/ So the first device with Convergence will really be a phone? [15:48] sturmflut2, yes [15:49] cheers sturmflut2 [15:49] any idea why QGSettings::set would cause a segfault? (parameters are correct) [15:50] it looks like something goes wrong inside qgsettings logic, but gdb doesn't want to cooperate atm [15:51] ogra_: Well, I suppose there are some SoCs available on the market which are really powerful enough to run a full desktop [15:51] sturmflut2, sure, no idea what the HW specs will be, i would expect a small netboot setup [15:51] *netbook [15:53] hey we had a full desktop running on n4 a few years ago, you'd be surprised how well it worked I think (and thats with a pretty old soc) [15:54] yeah, but not for day to day work [15:54] you want something more powerful and definitely more ram [15:55] (i guess the CPU would actually be okayish even, but for running a desktop (at the same time as your mobile UI) you want enough spare ram to still run apps like LibO) [15:56] cwayne: Probably depends on the type of application. I can imagine that window management and some graphics stuff is actually quite fast, because you can offload lots of things to the GPU, but all ARM SoCs out there are extremely limited when it comes to memory. I never understood why they put eight or more cores in a SoC, but at the same time can't even add a second memory controller, and some SoCs still don't even support [15:56] DDR3 memory at "regular" speeds [15:57] ogra_: we had a few people using it for day-to-day work actually (a couple people IIRC) [15:57] but yeah, it could be so much better with more ram and better cpu's now :) [15:57] just saying it's not that much of a stretch === tvoss is now known as tvoss|dinner [16:08] Even the MT6592 in the MX4 only has a single, 32-Bit wide DDR3 channel at a maximum clock of 667 MHz. Shared between GPU and CPU. That's less than half the memory bandwidth most netbooks have, and those are usually barely usable as a desktop. === fgimenez_ is now known as fgimenez [16:09] A full-blown Firefox or LibreOffice might be less than happy with that. [16:09] But it forces people to optimize their stuff ;) [16:11] cking: I did some tests, is it correct to assume that smemstat will also count the code segment as "shared" if the same binary is running multiple times? [16:12] sturmflut2, that is how I interpreted the shared mem stats that I get from the /proc interfaces [16:13] note that these values are instantaneous and are subject to variability [16:13]