[06:10] good morning [06:11] o/ jibel [06:12] hi tsimonq2 [06:15] good morning desktoppers [06:20] Morning jibel, tsimonq2, oSoMoN [06:24] o/ oSoMoN and duflu [06:24] hey duflu, tsimonq2, jibel [07:21] good morning [07:22] o/ didrocks [07:24] hey tsimonq2 [07:29] good morning desktopers [07:32] re seb128 [07:32] o/ seb128 [07:58] morning all [07:59] o/ willcooke [07:59] hey willcooke [08:01] morning [08:02] hey hey Laney [08:03] hey didrocks [08:03] good weekend? [08:05] quite uneventful, we wanted to see some museum for the European Heritage Days, but it started to rain when we went out [08:05] so we decided to head back [08:05] and you? [08:13] o/ Laney [08:13] oooooooooh what did I do [08:13] went to the theatre(!) on saturday, and yesterday it was the "green festival" which was fun [08:13] then to the pub in the evening [08:13] hi tsimonq2, how's it going? [08:14] aren't you in the US? [08:14] get to bed! [08:14] Laney: I procrastinated on $large-sum-of-homework :P [08:15] Otherwise I'm pretty good [08:15] How about you? [08:21] yeah awesome [08:22] it's a new dawn it's a new day it's a new life for me [08:22] and i'm feeling gooooooooooooood [08:22] :D [08:58] willcooke: hey I might have an idea on how to workaround that issue with the lockscreen [08:58] the only problem is that I cannot reproduce it [08:59] do you know anyone that can easily reproduce it? [08:59] so I can ask him to test the fix (as soon as it's ready) [09:00] andyrock, nice one, but no, I don't know anyone who can reproduce it. Seb said he could but only sometimes I think [09:00] ok I'll try to find a way to force it on my system [10:05] didrocks, is there a reason that the Ubuntu logo looks fuzzy/filtered in GDM, but crystal clear in Plymouth? [10:06] Ideally they should look the same and be on the same spot in both screens. [10:08] amano: yeah, I don't really know if it's a general rendering issue (which is making fonts blurry as well under wayland) [10:08] amano: it's exactly the same image [10:08] like, referencing the same path on disk [10:09] Ok. I was just wondering. [10:11] But you can consider lowering “Ubuntu“ on the Plymouth screen if that can be done easily. To have it on the same spot like in GDM. To have it more seamless. Maybe for 18.04? [10:12] Also, the GDM purple background is textured, unlike Plythouth [10:12] I think having it centered on plymouth makes sense [10:12] duflu: right, and I wonder if it's not the noise texture vs logo rendering like this [10:12] (which would be weird) [10:12] didrocks, just the noise texture I guess [10:13] we should give it a look in NY, removing the noise texture and seeing how the logo renders [10:13] * duflu *shrugs* and goes to dinner [10:13] enjoy your dinner duflu :) [10:14] add it to the list :D [10:14] o/ [10:14] I would be interested to see how it looks, my gut feeling is that it will look a bit odd. [10:14] But it would be cool when it goes in to GDM and looks almost seamless [10:15] yes, removing the texture for background doesn't give great results [10:16] I just think the logo shouldn't be impacted by the texture, that's just what we should look after (if this is what makes it blurry) [10:16] Maybe somebody can make a mockup with Ubuntu lowered. If it looks ok, then it doesn't matter that this placement doesn't make sense. [10:16] same kind of "things to fix" like blurry fonts on wayland IMHO (there is a bug about it) [10:16] amano: volounteering? :) [10:17] Sure ;) [10:18] Firing up Gimp in the evening. [10:18] sweet [10:18] (I guess not for 17.10 anyway) [10:18] Not that important anyway. A papercut bug ;) [10:25] Can Robert Ancell propose LightDM as an official successor to GDM to upstream GNOME? [10:26] The missing lockscreen bits for GNOME Shell would have to be added. And maybe that takes a year or so [10:28] But considering the security review in the MIR even GNOME would probably be happy to default to something saner. [10:30] he did that some years back (like 2011/2012?) [10:30] It is written by a GNOMER (Ancell), it is saner, it is sleaker, Ubuntu is getting closer to GNOME again. Maybe they would take it under the GNOME umbrella. [10:31] Maybe he should propose it again. My heart bleeds every time I click onto my GDM login name. [10:32] At the worst they say no again. [10:36] * oSoMoN yells at computer that logs out randomly when he presses Ctrl+C in a terminal [10:36] O__o [10:38] bug #1710637 [10:38] bug 1710637 in mutter (Ubuntu) "Input falls through to gdm3 and terminates the session on Ctrl+C" [High,Incomplete] https://launchpad.net/bugs/1710637 [10:38] yes, it started happening to me last week, and it's been plaguing my work days since then [10:38] what do you do to trigger it? [10:39] not 100% reproducible, that's the gotcha, but usually Ctrl+C in a terminal to clear the prompt triggers it [10:40] and I *think* sometimes other key combinations do it too, but haven't managed to figure out which [10:40] jdstrand: that could be the issue we discussed about on Friday ^ [10:40] something else happened to make ctrl-c behave like that [10:42] please try to figure out what it is, that's the key to fixing this [10:42] oSoMoN, you're on wayland ? it never happened to me on X always wayland. [10:43] But why would Ctrl + C on GDM3 cancel the session anyways? [10:44] * tsimonq2 is curious to see what comes out of that bug [10:44] jibel, yes, wayland [10:44] you can configure the active console so that key presses go to it instead of the display server [10:45] run sudo setupcon --force -k and you'll probably see the same bug happen [10:45] Laney, yes, I’ll pay close attention to what might trigger it [10:46] I think I saw a comemnt somewhere where somebody said that they saw it after doing something with snap ... [10:46] * Laney coughs [10:47] it is not only ctrl+c for alt-arrow or alt-Fn. In my case alt-f4 to close a window switches to tty4 instead [10:47] s/for/but/ [10:47] yes [10:47] please try to find out what makes it happen [10:47] * Laney hasn't seen it since fixing the previous instance (console-setup) [10:48] Laney, thanks for the gdm fix, I just finished the upgrade tests and always get the right session [10:48] nice [10:48] that fixes the 'default' session case [10:48] where you haven't used the cog menu to select anything before [10:51] Laney, and also the right session is still selected after a logout and you don't have to switch between entries twice to select a session [10:55] logout just happened again. I had typed in a gnome-terminal window "ls /path/to/some/file" using tab completion to verify the exact name of that file, then pressed Ctrl+C to get a clean prompt, and voilà, back to session login screen [10:55] Laney, jibel: are there some logs I can collect somewhere? [10:55] oSoMoN, the journal, I think that's all [10:56] the ctrl-c is a symptom of the bug, not the cause [10:57] we need to know where the earlier mode switch happened [10:57] yeah and journalctl will contain the entire session with maybe some clue [10:57] might do [10:58] I'm just saying, don't focus on what you were doing when you pressed ctrl-c [10:58] that's not the smoking gun - something else is [10:58] * oSoMoN inspects journal [10:59] is there a way to query some vt config or something to know if the system is in that state? [11:00] like to pull the status regularly and get a notification when things change [11:00] to have a clue what action was done around that time [11:00] don't know [11:00] if it's setupcon you could add something to that file to log when it's run [11:00] that one is a script [11:45] I worked out how to query it, I think, and attached a program to bug #1710637 [11:45] bug 1710637 in mutter (Ubuntu) "Input falls through to gdm3 and terminates the session on Ctrl+C" [High,Incomplete] https://launchpad.net/bugs/1710637 [11:48] oSoMoN, jibel ^ [11:48] Laney, good job! [11:51] ty [11:54] Laney, awesome, I’ll use that [11:54] make it notify you or something [11:55] well, this is all assuming that we're barking up the right tree [11:55] didrocks: oh, re bug #1710637. I commented in https://bugzilla.gnome.org/show_bug.cgi?id=772476 on that a long time ago, but I couldn't ever find a reproducer so didn't do as upstream asked [11:55] yes, of course [11:55] Gnome bug 772476 in wayland "[wayland] sometimes all keyboard input is displayed on vt1 when shutting down" [Major,New] [11:55] bug 1710637 in mutter (Ubuntu) "Input falls through to gdm3 and terminates the session on Ctrl+C" [High,Incomplete] https://launchpad.net/bugs/1710637 [11:55] thanks, i'll try it [11:55] K_OFF currently, presumably I’m in a good state, let's see if I can trigger a state change [11:56] didrocks: that is different than what I was seeing last week. this is different from what I saw last week-- I had mutter segfaults [11:56] muhahah, they ran my trojan [11:57] Laney, should I have snapped it to run it fully confined? ;) [12:00] man, that was fast [12:00] "snap remove libreoffice" triggered state change to K_UNICODE [12:00] and sure enough, the next Ctrl+C logged me out [12:01] SNAPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPPYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY [12:01] at least you can try 'FIX' mode next time :P [12:02] yeah, I'll keep your trojan handy until the bug is fixed :) [12:03] oSoMoN, write on the snapcraft forum about the issue [12:03] +1 [12:03] TODO: let didrocks know that after uninstalling gnome-shell-extension-dashtodock (I didn't use it), gnome-shell-extensions (don't think I used anything there), gnome-shell-extension-log-out-button (was using it) and gnome-shell-extension-suspend-button (was using it), the power button works as advertised [12:03] s/TODO: // [12:03] * jdstrand checkcs that off his list :) [12:03] seb128, doing that right now [12:03] jdstrand, how subtle :) [12:03] indeed :) [12:04] irc is my todo list backend now [12:04] * Laney gets the libreoffice snap too [12:04] suspend button is buggy with 3.26 it seems [12:04] I had the issue as well [12:08] ogra_: here? [12:08] ogra_: anything we need to fix in snapd/ [12:08] zyga-ubuntu, yeah ... seems the libreoffice snap somehow influences ctrl-c [12:08] no idea :) [12:08] ogra_: I don't have that snap [12:08] ogra_: but I have plenty of snaps [12:08] zyga-ubuntu, oSoMoN carries it to the forum this moment [12:09] ogra_: and I am affected by that [12:09] ogra_: I constantly kill gnome shell with ctrl-d or similar [12:09] it's unlikely having to do with the snap [12:09] (I started muscle-memory-learn to type exit instead) [12:11] I can reproduce with just snap install gimp [12:11] not me :-o [12:11] then alt-f4 swtiches to vt4 [12:11] let me reboot and try again [12:12] https://forum.snapcraft.io/t/snap-remove-somesnap-triggers-keyboard-input-fallthrough-to-vt-ctrl-c-in-terminal-logs-out-of-current-gdm-session/2162 [12:12] thats not a description, thats a novel :P [12:13] jibel: note that when snaps are removed we reload udev profiles [12:13] jibel: could that cause this? [12:13] jibel: I can tell you about other things we do [12:13] jibel: feel free to ask [12:13] ogra_, part II coming soon to your favourite book store [12:13] haha [12:14] zyga-ubuntu, perhaps you should do that reload a little more fine grained and exclude stuff like input devices (or harddisks you dont want these to change either underneath you) [12:15] udevadm trigger/settle has sepcific options for that [12:15] *specific [12:15] oSoMoN: I replied to the forum [12:17] ack [12:17] and again [12:18] zyga-ubuntu, under wayland open a terminal and run: snap install gimp [12:18] zyga-ubuntu, wait until it's install then press alt+f4 [12:18] doesn't happen for me on wayland [12:18] * Laney has super powers [12:18] checking [12:18] * zyga-ubuntu loves his network :) [12:18] if you cannot reproduce tell me what info you need [12:19] jibel: I pressed it, moved me to Ta [12:19] to VT [12:19] so I suspect it is broken now [12:19] Laney, don't you have a trojan of your own installed on your own system ;) [12:21] darn [12:21] I hit ctrl-c [12:21] haha [12:22] zyga-ubuntu: I put a program here https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1710637/comments/11https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1710637/comments/11 that you can use to fix it when it breaks [12:22] Ubuntu bug 1710637 in mutter (Ubuntu) "Input falls through to gdm3 and terminates the session on Ctrl+C" [High,Incomplete] [12:22] mvo: hey, so to avoid risking any more shell terminations, can you please merge https://github.com/snapcore/snapd/pull/3932 [12:22] Laney: I saw, I'm getting it now [12:23] sooo strace snapd and see if it calls this ioctl? [12:23] then look at the things that it calls [12:23] Laney: which ioctl? [12:23] Laney: I commented on the forum about what we do [12:24] ok, but someone who can make it happen needs to look at those things [12:24] https://forum.snapcraft.io/t/snap-remove-somesnap-triggers-keyboard-input-fallthrough-to-vt-ctrl-c-in-terminal-logs-out-of-current-gdm-session/2162/2?u=zyga [12:24] KDSKBMODE [12:24] Laney: after the crash I get K_OFF now [12:25] Laney: snapd doesn't do that for sure [12:25] didrocks: the very first commit after 3.26.0 looks interesting to cherry-pick https://git.gnome.org/browse/gnome-shell/log maybe it could close LP: #1713313 [12:25] Launchpad bug 1713313 in xdiagnose (Ubuntu) "Unable to launch pkexec'ed applications on Wayland session" [Undecided,Confirmed] https://launchpad.net/bugs/1713313 [12:25] Laney: but udev might [12:25] zyga-ubuntu, it does not happen with every snaps, for example 'hello' doesn't break the system [12:25] jibel: I suspec it is related to udev rules [12:25] please could you debug it? [12:25] jibel: libreoffice probably uses opengl [12:25] jibel: so you get udev rules [12:26] we generate the following udev config for gimp: [12:26] SUBSYSTEM=="drm", KERNEL=="card[0-9]*", TAG+="snap_gimp_gimp" [12:26] KERNEL=="nvidia*", TAG+="snap_gimp_gimp" [12:26] KERNEL=="vchiq", TAG+="snap_gimp_gimp" [12:28] zyga-ubuntu: hm? how is 3932 related to shell terminations? [12:29] mvo: it's not, I was just afraid I'll kit ctrl-c again and kill shell again [12:29] *hit [12:36] jbicha, why so much hate against system-config-printer? ;- [12:36] ;-) [12:43] I'm just cleaning up :) [12:44] although I would have been unhappy to have it used instead of the upstream Printers panel at this point [12:45] it's an easy clean up since it's not even visible in GNOME [12:46] jbicha, I'm not convinced it wouldn't be a good idea to keep it as a power tools [12:46] the printer panel is quite buggy and has some limitations [12:47] power tools -> solution for people who need things not working in the panel [12:47] like sharing their printer [12:47] or did they fix that in 3.26 ? [12:47] printer sharing is not in 3.26 [12:47] :-( [12:48] well that would be a good reason for me to keep s-c-p installed (and remove the onlyshowin) [12:50] wouldn't it be a bit confusing to have 2 things named Printers? [12:50] no Till around to ask him [12:51] jbicha, you want to argue in favor or removing the buggy & incomplete integrated panel right? :p [12:51] seb128, just got an email from him, so he'll be here soon I think [12:51] willcooke, thanks [13:11] when I try to type in firefox certain keys don't work now... like "a" (it works elsewhere)... man, what a day [13:16] weird [13:16] probably a snapd bug :p [13:25] zyga-ubuntu, it seems that all the snaps that trigger the bug use the opengl interface [13:26] jibel, does "udevadm control --reload-rules" or "udevadm trigger" trigger the same issue? [13:26] seb128, no they don't [13:27] jibel: thank you, that's a good hint [13:28] zyga-ubuntu, and it's only under wayland [13:40] seb128: could you try to restart the desktop versions tracker? I fixed a typo yesterday that broke it [13:40] k [13:52] seb128, udevadm trigger does actually trigger the bug under wayland. I added 2 test cases to the report. [13:53] first test is a high level test of the second [14:10] jbicha, RE cleaning up: vino and remmina do not work with wayland. will those be removed from main? [14:12] jibel: nice, probably a systemd bug then ;-) [14:12] * desktop team's sloping shoulders [14:12] https://lists.ubuntu.com/archives/ubuntu-desktop/2017-June/004981.html [14:12] amano_: ^ I asked about that earlier but at this point it would also need a UIFe in my opinion [14:18] can you still drop the xdiagnose dependence on ubuntu-desktop, so that can be removed manually (without removing the meta package)? [14:18] s/dependence/dependency [14:27] amano_: I think we need more information from tjaalton about what the consequences are of dropping xdiagnose for the failsafex feature [14:29] is anyone using artful and owning an internet stick to test bug #1707023? [14:29] bug 1707023 in gnome-control-center (Ubuntu) "mobile network connection cannot be added" [Undecided,New] https://launchpad.net/bugs/1707023 [14:29] jbicha: failsafe-x can go [14:29] willcooke, ^you maybe? [14:29] I think cyphermox has those [14:29] tjaalton: so there's no objection to dropping xdiagnose from default install? what about xterm? [14:30] jbicha: what about xterm? [14:31] I don't need that [14:31] tjaalton: my understanding is the only reason that we've had xterm in the default install for years is because of failsafex [14:31] seb128, Let me see if I've got a SIM card that will work [14:31] jbicha: ok, so it can go as well then [14:32] willcooke, thanks [14:32] weird if xdiagnose didn't depend on it directly then [14:39] i wish ImageMagick could at least be hidden in gnome-shell Activities and gnome-software [14:39] the icon is off on so many levels [14:39] We have a list of things to put in "Utlities" section in g-s [14:39] Utilities* [14:40] and removing it from gnome-software removes cups. [14:40] why do we need imagemagick again? [14:40] I guess we can work on that in NY, just going over them one after another to minimize the number of default entries in the overview [14:40] I guess that's it :p [14:41] printing. part of it is used for printing. but that isn't obvious to the end user. It shouldn't show up in gnome-software [14:42] see the final comment at https://bugs.debian.org/675617 [14:42] Debian bug 675617 in imagemagick "Please hide the menu entry" [Wishlist,Open] [14:42] too bad the Debian maintainer isn't that interested yet but we'll probably make that change for Ubuntu anyway [14:42] right [14:42] jbicha: IIRC, willcooke was against hiding it as well, but moving it to the Utilities category rather [14:43] don't think splitting up helps? quite a divergence from debian… [14:43] didrocks: my proposal doesn't hide it, it just doesn't install that piece by default but it's available to install for the few people who do want it === JanC_ is now known as JanC [14:43] yeah, it's a large change and maintenance overhead, but I don't have strong opinion either way :) [14:43] we don't intentionally want the display app in the default install, it's a packaging mistake in my opinion [14:43] (this UI is horrible btw) [14:44] exactly [14:44] we can't support that app [14:44] no app that removes big parts of the software stack should show up in software. imagemagick removes the printing support, removing xdiagnose removes the ubuntu-desktop meta package [14:44] I'm sure it's following the HIG, at least, the one from GNOME 1 ;) [14:44] jbicha: do you want to tackle the split? [14:45] yes, I'll file a bug with my proposed patch for review [14:45] nice! [14:45] I think that was already on my todo list but I blame Irma [14:45] there are others we want to keep installed by default, but they should just go to Utilities in that case [14:46] (you can find back the list on the fit and finish sprint trello card, it's one of the item when we reviewed the list) [14:46] amano_: the GNOME/Ubuntu Software app allows marking apps as required for a particular desktop. You can see those in the Installed tab as "System Applications" [14:46] they cannot be removed using the GNOME Software app [14:47] so we should use that feature to match the ubuntu-desktop depends. Not sure if that will happen this cycle though. [14:48] I love how vim always shows up as installed on a fresh install in gnome-software ;) And apt is pretty sure that it isn't [14:49] amano_: see this patch: https://anonscm.debian.org/viewvc/pkg-gnome/desktop/unstable/epiphany-browser/debian/patches/dont-make-compulsory.patch?view=markup [14:49] it'd probably be better to ship some appstream xml centrally to do that [14:54] seb128, having problems getting my 3G dongle to show up at all in Xenial on U7 where I /think/ it used to work. Could be a bigger problem, could be just me [14:55] willcooke, ok, don't worry/bother, maybe cyphermox or somebody else can help with that or maybe we can find somebody with a dongle next week [14:55] usb_modeswitch seems to be doing it's thing, so could be the SIM card [14:56] seb128, also - hub post ;) [14:57] seb128, I'll do that [14:57] good idea! [14:59] willcooke: popey: idea for the hub, would be cool if there was a way to replace the comment system for some blog post by a hub discussion embeeded. That would centralize all discussions on things we post on (insight posts, a particular person blog post) on the hub [14:59] sorry, I don't have the tools to test this anymore [15:02] didrocks: like this? https://meta.discourse.org/t/embedding-discourse-comments-via-javascript/31963 [15:03] popey: sounds exactly it! [15:03] popey: I guess I'll need to find out how we can make that work with my static site generator, and somebody should do the same for insights :) [15:04] oooh! Good call. [15:04] seb128, I fixed it. *cough* By unplugging ir and plugging it back in a few times. [15:04] *shrug* [15:04] * didrocks urges himself to resist on pasting an IT Crowd reference [15:04] willcooke, lol [15:07] did you switch it off and on again? [15:08] :-D [15:14] jbicha, versions updated [15:17] thanks! [15:18] yw! [15:25] seb128, updated that bug - have got it working, but the control center panel is a bit lacking. Will review next week and then log upstream bugs [15:25] willcooke, thanks [16:14] cyphermox: just assigned that ctrl-c terminating the session bug to you, another instance of a similar thing in console-setup [16:14] it's debugged, just not sure what the correct fix is, maybe you can take a look please? [16:20] sure, but I thought this had already been fixed [16:20] another instance [16:21] ok, what bug # ? [16:21] https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1710637 [16:21] Ubuntu bug 1710637 in console-setup (Ubuntu) "Input falls through to gdm3 and terminates the session on Ctrl+C after udevadm trigger is executed under wayland" [High,Triaged] [16:22] probably systemd does this properly these days anyway so that rule isn't that important? [16:39] (commented on the bug, but I just tested and it seems like that is true) [18:46] Woop woop? Is there any improvement to IOMMU support in the latest iteration of Ubuntu desktop? [18:49] ThorHop[m]: you could try booting with intel_iommu=on or intel_iommu=strict and find out if your drivers are any good.. [19:08] sarnold: according to qubes website my dell xps 9530 supports IOMMU... with an exception of booting from external harddrive. So I'll have to install on my SSD :P [19:10] and older kernel, without the latest nouveu drivers... hmm... === Guest41832 is now known as RAOF