[00:40] the moment after upgrade... and having to purge nvidia driver [01:27] Trevinho, welcome to the cool timezone [01:40] the quiet timezone :) === brainwash_ is now known as brainwash [02:05] jbicha: Indeed. :) [02:06] Tomorrow comes today! [02:08] [TM] [02:27] robert_ancell: well... It has some good points, yeah.... [02:38] Judging by the understatement, I imagine Trevinho is actually still in Italy and not enjoying the other good points of this timezone in person :) [02:55] duflu: no.. I'm in Bali.. 😂 [02:55] Trevinho: Oh quite nearby then :) [02:56] Trevinho: Enjoy! [02:56] And robert_ancell, being always been a night owl, this allows me to enjoy more the day and work with others later... Never though about it! 😊 [02:57] duflu: Yeah.. I should have come there for some days too... But I didn't plan it in time 😕 [02:57] Trevinho: Other stops in asia? [02:59] duflu: Singapore, and before here Kuala Lumpur [02:59] Just this area though [03:00] Trevinho: I'm glad. IMHO you wouldn't want to go to the other side of the planet just for Bali. [03:00] Although some Europeans do [03:01] duflu: well I'm staying here for a month, so... Then I had to come back... [03:01] Well, so I thought, but it was in fact not a so great decision. [03:03] Trevinho: I'm not really sure what you mean but if you wanted to you could be in Australia before dinner :) [03:05] That said, seems like winter is creeping in about now. I'm sure it's much warmer and the cocktails are cheaper in Bali [03:06] duflu: Yeah... Having lunch and dinner for 2€ (each) is kind of cool. And temperatures are perfect === JanC is now known as Guest52949 === JanC_ is now known as JanC === brainwash_ is now known as brainwash [06:52] so good morning everyone [06:56] Morning allison[m] [07:01] hey duflu [07:01] how are things? [07:03] moin [07:03] allison[m], things are confused but have been worse plenty of times. "Good" would be accurate. You? [07:05] pretty good, i think [07:06] Good morning [07:06] Moring pitti [07:06] good morning, pitti [07:06] Morning too [07:06] so, big announcement, eh! back to GNOME.. [07:06] hi RAOF [07:06] Good morning pitti and allison! [07:06] hey duflu, RAOF, and allison[m], how are you? [07:06] (this moment in time travel, brought to you by the matrix) [07:06] allison: Greetings, fellow Matrixite :) [07:07] seb128, didrocks: bonojur mes amis ! [07:07] pitti: ya... no kidding! [07:07] bonjour pitti, allison[m], Sweetshark, RAOF ! re seb128 [07:07] I shed a tear for unity7, I like that much better than shell [07:07] * RAOF blames allison :P [07:07] and duflu ;) [07:07] my 6-years-long policy of unity subversion has finally paid off [07:07] I've actually been using shell for the last month on half of my machines. [07:07] but I think binning mir and u8 is the right strategic decision [07:07] cough [07:08] Hey didrocks [07:08] It's grown much nicer, and with extensions you can get most of the features I used to miss from U7. [07:08] good morning desktopers [07:08] RAOF: only after I tried it I realized how much chrome (aka wasted space) that has.. [07:08] ya. to be honest, i'm a bit surprised about unity getting the axe, and i sort of hope we at least make some attempts to do some UI chnages to shell to be more unity-like... [07:08] but as for mir... shrug no big loss, i think [07:08] hey didrocks allison[m] pitti RAOF duflu [07:08] but, happy i3 user here [07:08] pitti: Yeah, there is that. [07:08] Morning seb128 pitti allison[m] [07:08] Hey seb128! [07:09] Morning seb128 [07:09] allison[m], is there any way you drop that [m] btw? :p it somewhat make some graphical clutter I find... [07:09] seb128: give me a week or two =) [07:09] lol [07:09] k [07:09] * allison[m] is waiting for a nick to expire in nickserv =) [07:09] Lack of locally-integrated-menus (and, partially the HUD) are the bits that still grate. [07:10] You can claim that back, you know :P [07:10] ya. i know. but only after 15w of idle :) [07:10] why not allison_ meanwhile then? [07:10] allison[m], what uses 15W? [07:10] Is your freenode account also without an associated email? Sucks. [07:11] my point is that the nick belongs to someone else :p [07:11] Oh. That's somewhat different! [07:11] allison[m]: hope they don't appear in 14 weeks and 6 days then :D [07:11] no joke :) [07:12] At least you own the matrix name :) [07:12] Oh, possibly U8 uses 15W (bug 1664828) [07:12] bug 1664828 in unity8 (Ubuntu) "Unity8 wakes up continuously at around 62Hz on an idle desktop" [Medium,New] https://launchpad.net/bugs/1664828 [07:13] ya. pretty happy to have gotten the matrix name. that was quite impressive. [07:17] Or possibly ominous. Are there no women on matrix at all? :X [07:18] A sadly plausible possibility, given the sausage-fest that's common in FOSS communities (that aren't explicitly trying to be not sausage-fests). [07:18] a friend failed to register her (far less common) name [07:19] i think i just got really lucky [07:19] :D [07:20] * RAOF did manage to claim his rightful RAOF@mastodon.social and RAOF@icosahedron.website accounts. [07:21] Which you'd think would be pretty trivial, but there's some bastard who keeps going out and registering RAOF in lots of places! [07:21] I still have no clue what matrix looks like and why people find it so hype [07:21] need to go to some conf or hackfest to have people demoing it [07:23] found you :) [07:23] hello everyone [07:23] Morning hikiko[m] [07:24] hi duflu [07:24] I was exploring matrix [07:29] not bad tbh [07:31] better than your old IRC client? in which way? [07:32] not better [07:32] just not bad [07:32] matrix is supposed to have end to end encryption in private chats [07:32] appart from that I can't see any difference [07:33] I was talking to some of my friends about matrix the other night. They've setup server and integrate bridges to link different protocols. [07:33] +I prefer IRC because I have already setup znc and connect there from anywhere :p (desrt is gonna say how graybear is that now :P) [07:33] They could say enough good things about it. Except the UI is butt ugly. [07:35] the dark ui is ok... well my favourite client is irssi but I use hexchat for work to not miss any notifications [07:45] Hey folks. :) [07:52] hi TheMuso [07:56] Morning all. Much happen while I was off? [07:56] Hey willcooke. :) [07:57] hi willcooke :) [07:57] Morning TheMuso, how goes? You got my message about skipping 1:1 today right? [07:57] willcooke: Indeed I did, no worries. [07:57] Hope you didnt hang around late [07:57] cool [07:57] they hikiko[m] [07:57] No not around for work purposes really, but still around anyways. [07:58] Since the daylight savings changes our meetings are earlier for me now anyways. [08:01] morning you subjects of the news [08:02] hey willcooke, morning Laney! [08:02] haha [08:03] Laney: Morning fellow subject of HRH QEII [08:03] I think [08:03] correct [08:03] * Laney looks up at the portrait on the wall [08:03] light a candle below it [08:04] I assume you mean not right at the bottom [08:04] haha [08:05] Although that would be a fairly common point of view being expressed [08:05] testing if god saves the queen portrait [08:05] Heh [08:05] willcooke: good morning [08:07] Wow, only queen of four countries. I thought it was more [08:07] depends on your definition of "country" i guess [08:08] vs commonwealth [08:08] i mean, her record is pretty sad [08:09] countries: 12 lost, 4 remain.... with 1 looking to be in pretty bad shape at the moment [08:10] Bad shape is relative. Can you really say that with the state of Syria? Yemen?... [08:14] ...heavy [08:14] If the city of london and scotland and ireland remained in EU then the shape would not be too bad: it would be the shape of a donut :p [08:14] don't forget manchester.... [08:14] (/me says bad jokes) [08:16] anyway I think brexit = just a different agreement between UK and the rest of EU and nothing more, I wouldn't worry so much [08:16] hey willcooke Laney, how are you? [08:17] hey seb128 [08:17] alright, got shorts on today! [08:17] although that might be ambitious [08:17] it's colder today here [08:17] was like 17°C on sunday [08:18] and we are back to 7°C this morning [08:18] can't resist a blue sky [08:18] it's like 9 now [08:18] Finally mornings are cooler here too, around 11C the last couple of mornings. [08:18] Bring on winter. :) [08:18] went out to the french restaurant last night after reading THE announcement [08:18] was good [08:19] hey TheMuso [08:19] Hey Laney. [08:19] seb128: TheMuso: how's it going? [08:19] That would have been nice indeed. [08:19] Laney: Honestly, really happy right now. [08:21] Laney, good! sounds like quite some happened around here just after I left for tennis yesterday [08:21] Laney, french restaurant, getting used to the idea that if we go back to things we used to do then french by default should be considered again? ;-) [08:23] * larsu looks into ubuntu-desktop after a while of absence, still talking about French as a default [08:23] all good! [08:23] lol [08:23] ...seb sees his chance [08:23] * didrocks updates the seeds for zesty [08:23] 1) wait for confusion to erupt [08:23] 2) seize power [08:23] hey larsu :-) [08:23] indeed, a lot of seed changes going to happen [08:23] would be easy for fingers to slip... [08:23] haha [08:23] larsu: "oupsss" [08:23] :) [08:24] hey seb128! [08:24] and everyone [08:24] hey larsu... you're missing the party! [08:24] ahoy larsu [08:24] there's a party? [08:24] you didn't need to go to RH. we're switching to gnome here now :) [08:24] Laney: trimming down disk size again… "But I had to drop -en…" [08:24] haha yes! [08:24] did you see that cockpit is in zesty? [08:24] I did! [08:25] yep, I even heard the guy who did it talk about it ;) [08:25] pitti: yeah, the packaging SUCKS! [08:25] * larsu high fives pitti [08:25] cough [08:25] and I just uploaded 137-1 to clean it up (arm FTBFS, autopkgtest failure, cockpit-ws not run as root any more) [08:25] I was a bit surprised to see it in zesty already, I aimed for aadvark [08:25] but, so be it [08:25] https://github.com/cockpit-project/cockpit/pull/6288 :) [08:26] 138 will actually be a nice version to support in a stable release [08:26] now we can say that it's been in there forever, since *before* the alphabet wrapped around [08:26] there are some ... [08:26] keen developers [08:27] now the last step is just to get it into xenial-backports :) [08:27] we do upstream CI on xenial, so that's the real target [08:27] imho simply allowing the alphabet to wrap shows a lack of imagination [08:27] we have unicode.... lots more letters to explore... [08:28] do what excel does [08:28] aa, ab, ac [08:28] there's still 'c' and 'a' left [08:28] aaudacious aardvark [08:28] let's use up the Latin alphabet first before moving to a new one :) [08:28] aawkwaard [08:28] IMO go back to A, and make this next cycle 12 months instead of 6. :p [08:29] aaliterative aardvark would be better [08:34] Trevinho: hey! re-poke regarding https://bugs.launchpad.net/ubuntu/+source/bamf/+bug/1680104 [08:34] Ubuntu bug 1680104 in bamf (Ubuntu) "Package FTBFS" [High,In progress] [08:35] Trevinho: as mentioned in the bug, I can fix this by just adding a -Wno-deprecated-declarations but I would prefer to properly fix it ;p [08:49] -Werror=deprecated-declarations in release builds sucks [08:53] indeed [09:08] you can always plit ireland up in more countries if you care about the counter ... [09:08] s/plit/split/ [09:13] aardvark wouldnt even scale one step beyond aardvard: bb.* ? [09:16] bbq of course! ;) [09:19] RAOF: still around? what gnome-shell extensions did you enable to get it closer to behaving like u7? [09:19] haha [09:19] just use xenial if that's what you want [09:20] tjaalton: primarily the sound menu one; that's the one I miss the most. [09:21] RAOF: ah, alright. I'll give it a shot, thanks [09:22] Specifically: https://extensions.gnome.org/extension/55/media-player-indicator/ [09:26] yup [09:40] Laney: given yesterday's announcement, I guess the desktop-team party was big ^^ [09:41] hey ximion [09:41] maybe Canonical could indeed become a counterweight to Red Hat in GNOME (IMHO that would only help the project) [09:41] spare a thought for those working on u8 stuff [09:41] I was reading Phoronix yesterday and given the talk there, nobody in the u8 team knew about it... [09:42] (and yeah, not the most reliable source) [09:44] no comment :) [09:45] some interesting opportunities ahead for sure [09:46] it's really sad for the phone stack [09:46] on a different note, do you have any idea on whether you plan to use GNOMEs / GNOME Software's "web-application" feature? [09:47] never heard of it [09:47] it's basically AppStream data with a hyperlink in there which gets opened in an Epiphany window with less chrome [09:47] oh is it epiphany in chromeless mode? [09:47] jup [09:47] but with predefined stuff in GNOME Software [09:48] it's a rather weird thing and personally I don't see much use for it, but GNOME likes it :P [09:48] can't see us shipping epiphany by default [09:48] ok [09:48] but if it could work with firefox then sure? [09:50] Laney: I'll talk to hughsie about it :-) - will probably not be something for the AppStream 0.11 release, that release has already too many new features enqueued [09:50] where did the idea come from? [09:50] how does it even work? [09:50] you get a package with a metainfo file? [09:51] you just get a metadata snippet in the AppStream file that doesn't belong to any package [09:51] how does it get in there? [09:51] I think Fedora injects those at the generator level [09:51] interesting [09:52] as in "take this extra metadata and put it in the result" [09:52] so yeah they can be nice [09:52] just don't know if we'll be able to run them in such a good way [09:53] I guess in g-s 'installing' them makes a desktop file or something [09:53] yes [09:53] with some special Epiphany arguments [09:53] maybe the webbrowser app can live on ... [09:54] it's pretty tied to Epiphany and there is no clear definition on what a web application actually is, that's why this feature isn't in AppStream yet, but I am not really opposed to having it [09:54] not sure this isn't scope creep though [09:54] now appstream is a website description format too? [09:55] a "web application description" apparently a webapp is something special :P [09:55] if you say that they're apps then maybe it makes sense [09:55] I suppose they are app-like once installed [09:55] yeah [09:55] web is weird [09:57] bloop [09:58] From an a11y perspective, I'd prefer firefox or epiphany, although firefox is better on the a11y front. === hikiko is now known as hikiko|ln [11:32] sil2100: I will fix it later ok [11:34] Trevinho: if you're busy currently would you be fine for me working-around it and then leaving it for you to fix it properly in the nearest future? [11:34] I want to clear as much as possible from the FTBFS list before release of zesty [11:47] sil2100: let me check a sec [11:48] sil2100: oh, never mind... yeah, please do that... As we don't really care much at this point of deprecated. [11:48] sil2100: even in unity we've that, so it's fine... As they won't last longer anyway. [12:09] umount --just-do-what-i-say [12:10] * Laney wins [12:11] ahaha [12:12] Yeah, it's always the most drat command so far [12:31] Trevinho: \o/ ok, thanks! === hikiko|ln is now known as hikiko [12:35] Fedora is odd in that they ship epiphany in the default install but without the .desktop so it looks like there is only one browser installed [12:36] I believe part of the justification was to make it easy to "ship" documentation for some of their projects by just putting the docs on a website [12:37] (and mcatanzaro would love for Fedora to switch to epiphany eventually) [12:37] anyway, I'm out for a bit… [13:11] seb128: can have you a look at that u-c-c branch, so at this point I guess we can try to deliver it to z and eventually SRU it. [13:11] Trevinho, I started but you have like autotools changes mixed in that for whatever reason [13:11] that's not appropriate for z [13:12] seb128: I inherited these from hikiko branch, not sure what it wasn't working to her [13:12] but you are right I should comment on the mp saying that [13:12] seb128: sure, but since there won't be an a with it, it's the only way.... If we've a FFe. [13:12] but usually changes should be at least explained [13:12] and different logical changesets should be in different mps [13:12] you got a ffe? [13:13] or, well, we can still push to a, and then use that as source for SRUs? [13:13] we can as well yes [13:13] hikiko: did the bug, but I guess it wasn't ack'ed [13:13] it feels a bit late to land a rewrite [13:13] you can't deliver this to z [13:13] it's a non LTS serie and it's not critical to land that this cycle [13:13] Ok, fine... [13:13] well, maybe as an SRU [13:13] which bug? [13:13] but that's an argument to have with them [13:14] I wasn't in rush, just at this point i don't want to loose these changes. As they can still go to X in some ways [13:14] the lowgfx switch [13:14] Trevinho, it's on my "to-review" list for this week don't worry [13:14] right, get them in z+1 then try to sru [13:14] Laney: you can still remove the switch... The rest of changes are bug fixes [13:14] seb128, Trevinho which bug are you talking about? [13:14] if you have a good plan for that it should be doable [13:14] as right now if you save settings to lowgfx they won't apply to normal profile [13:15] hikiko, the lowgfx u-c-c changes [13:15] hikiko, you have/had autotools file changed in there without explanation of why that's needed [13:15] oh [13:15] seb128: in the commits there was someting like "remove deprecation warnings" [13:15] seb128, I was having error [13:15] yep [13:15] error or warning? [13:16] can't remember to be honest, I suspect error otherwise I wouldn't have reason to change it [13:16] no big rush imho [13:16] seb128: I can try to rebuild it [13:16] Trevinho, k [13:16] but I could revert that [13:16] no, no rush... [13:16] it's not a big deak [13:16] it's not a big deal [13:16] hikiko: it's up to me, don't worry [13:17] hikiko: but you got the FFe, or just opened the bug? [13:17] hikiko, don't worry, but next time it's good to have a summary of why changes are needed and what they do in the commit msg or changelog [13:17] I only opened the bug I couldn't ask for a ffe without having my branches reviewed Trevinho [13:17] you could [13:17] sure seb128 you are right [13:18] hikiko: the branches are reviewed right now... I've redone the UCC work mostly as it needed some other changes, so that won't go in, the unity one also got some changes on my side, so... you're all set from code point of view. [13:19] seb128: hey, I suspect you've been asked this already, but do you know the general state of gnome3/wayland in zesty? wondering if it is worth it to develop wayland/gnome3 classic interfaces for snappy no or waiting til 17.10 opens [13:19] now* [13:19] jdstrand, hey, I don't think anyone asked but zesty should be similar to whatever we are going to have over the next cycles so it's a good base to work on that [13:20] so, next step is to contact the ubuntu-doc and ubuntu-translators teams? [13:21] I ve subscribed the bug to the ubuntu-release team [13:21] the ubuntu-release team to the bug* [13:21] :p [13:22] I am reading this guide: https://wiki.ubuntu.com/FreezeExceptionProcess [13:22] seb128: ok, cool. I saw the response from fedora on the status of gnome3 wrt the feedback received in kirkland's reddit and was curious if gnome3 in the archive was (nearly) as good as fedora's [13:23] jdstrand, I don't think fedora has much distro specific, they just hire upstream hackers to get work done upstream so we should be mostly on the same level [13:24] seb128: Makefile changes reverted though, it works anyway [13:24] Trevinho, thanks [13:25] hikiko: I think you should be going for z+1 and SRU rather than freeze exception [13:25] jdstrand, well some improvements might be from GNOME 3.26 and we are still on 3.24 for zesty [13:25] the final freeze is tonight [13:25] Laney: ok, that's fine for me [13:25] Trevinho: ya, just making sure everyone gets the message ;) [13:26] turns out this can of compressed air is very flammable [13:26] * Laney tried to put a candle out with it for fun [13:27] Laney: it is... I had some fun time ago with it [13:29] willcooke, ^ is what Laney says fine for you too? (for the low gfx) [13:30] Laney, I don't know how important the feature is to be honest but tonight seems to be very close :) [13:30] indeed [13:30] otp - gimme a sec [13:43] hikiko, what laney says is a good idea. Too late now [13:46] seb128: ok, thanks for the info [13:47] jdstrand, np, feel free to ask more questions on this channel if needed we are around and happy to help! [13:47] ok === Trevinho_ is now known as Trevinho [14:11] I switched to zenial/gnome yesterday, the alt-tab, alt-` is still there, so that's a nice bonus [14:22] jcastro: it's been years, but if I remember correctly that appeared in GNOME before Unity [14:22] jcastro: are you using Dash to Dock? [14:23] no [14:24] if you like a side launcher sort of like Unity it's a good choice [14:24] https://tracker.debian.org/pkg/gnome-shell-extension-dashtodock [14:24] I'm going to go as extension free as I can, I did load the weather applet one, which is lovely [14:25] there's a weather applet built into the GNOME 3.24 clock menu [14:26] (you need to have gnome-weather installed for it to show up) [14:27] willcooke: I like headline #4 at http://www.omgubuntu.co.uk/2017/04/5-questions-asking-future-ubuntu [14:27] "Will Happens to Snappy and Snap Apps" [14:27] :D [14:28] Those poor guys [14:31] jdstrand: Fedora 25 and GNOME 3.22 switched the default GNOME session to Wayland, Debian and Ubuntu GNOME currently still default to X but we've shipped the Wayland session since 16.04 LTS [14:33] http://blog.dustinkirkland.com/2017/04/thank-you-note-to-hackernews.html says 'It would be difficult, though not necessarily impossible, to offer the grsecurity supported in the Ubuntu archive' but surely not for technical reasons as debian has a linux-grsec package [14:33] probably the biggest downside is if Wayland crashes, it takes down your whole session (GNOME on X can restart gnome-shell when it crashes) [14:33] GDM3 though is actually Wayland by default though [14:34] jbicha: fyi, I just tried "GNOME Wayland' from lightdm in a vm and got a blackscreen. going to gdm3, it seems to go into wayland. /usr/lib/gdm3/gdm-wayland-session ... [14:35] jbicha: I see a bunch of Xwayland in there [14:35] I don't know if that is normal [14:35] femme: grsecurity is now for paying customers only [14:36] XWayland is always running; it would be a bunch of work to split that out to only run on demand [14:37] mdeslaur, I know, the patches to testing are still available though and there will probably be a community effort around maintaining them until the next public release [14:37] Firefox is still XWayland; chromium needs to be gtk3 first but that's happening later this year I believe [14:43] femme: doing that is a dead end though, forward porting old grsec patches is a bad idea [14:44] anyway [14:44] jdstrand: do you have time to file a bug about GNOME on Wayland being broken with lightdm [14:44] my understanding is that the reason gdm3 uses Wayland by default is because it's needed for the GNOME on Wayland session to work [14:45] so maybe for zesty we should see if we can blacklist that session for lightdm [14:45] why not just kill lightdm? [14:46] wouldn't it make sense just to go with gdm if we're going to be switching to gnome? [14:46] well lightdm is used by every flavor except Ubuntu GNOME and Kubuntu (sddm) [14:46] but gdm does integrate better with gnome-shell === maclin1 is now known as maclin [14:51] shell uses gdm for the lockscreen somehow doesn't it? [14:51] the reskilling starts already! :) [14:52] does gdm do guest sessions nowadays? [14:52] yes, I think it inspired the unity-greeter implementation :) [14:52] seb128: no, I don't think so [14:52] :-( [14:52] don't sad face [14:52] this is job security :P [14:52] lol [14:53] I think that's a good feature anyway [14:53] yeah me too, I use it a lot [14:53] I guess here start our list of dev work needed :p [14:53] jbicha: sure I can do that [14:54] jbicha: against what package? [14:54] Privacy is big so I can see a guest mode being a popular improvement to GNOME [14:54] ugh, i can't work without the guest session! [14:54] jdstrand: if we want to just blacklist it, maybe just lightdm; but the session files are from gnome-session [14:55] i'm still using lightdm, maybe i should switch to gdm [14:55] kenvandine: we need you to still work though!! :) [14:55] our guest mode relies on apparmor, so I'm not sure it's upstreamable as-is [14:55] :) [14:55] mdeslaur, worse than running vanilla linux? [14:55] guest session is super useful for testing [14:55] seb128: are there any snaps that you know of that will use wayland? I found gnome-easytag but it just crashes and I found glade-attente, but it needs X [14:56] (note, glade-attente uses Xwayland fine) [14:56] jdstrand, any GTK one? [14:56] GTK3 [14:56] femme: go read the grsecurity blog about the security issues introduces by people who forward-ported the patches [14:56] seb128: can you name one otoh? snap find gnome doesn't help [14:57] femme: anyway, point is if you want grsec, you need to pay for it now, that's how it is [14:57] jdstrand, I'm looking sec [14:58] mdeslaur, huh? do you mean the ones about KSPP? I'm just talking about maintaining the LTS which is the current plan [14:58] jdstrand: i can try re-building the glade snap with wayland support if you can't find anything else [14:58] mdeslaur, but even then, I'm afraid it's still an improvement [14:58] * femme is using grsec [14:59] attente, wdym with wayland support? shouldn't gtk just use the right backend depending of where you start the binary? [15:00] jdstrand, try https://uappexplorer.com/app/ghex-udt.canonical ? [15:00] jbicha: https://bugs.launchpad.net/ubuntu/+source/gnome-session/+bug/1680496 [15:00] Ubuntu bug 1680496 in gnome-session (Ubuntu) "black screen when using 'GNOME Wayland' with lightdm on zesty" [Undecided,New] [15:01] seb128: i think it doesn't have wayland support because i didn't add libwayland-dev to the list of build packages [15:01] so i assume it just doesn't have that backend built in that snap at all [15:01] attente, oh, you built gtk from source? [15:02] should be able to check that with ldd [15:02] attente: thanks for the offer. trying ghex [15:02] not side staging the deb? [15:02] seb128: yeah, that's a snap i built using jhbuild [15:02] so it has everything from the ground up pretty much [15:02] k [15:03] seb128: fyi, ghex is trying to access the X socket too [15:03] :-( [15:03] apparmor="DENIED" operation="connect" profile="snap.ghex-udt.ghex" pid=16352 comm="ghex" family="unix" sock_type="stream" protocol=0 requested_mask="send receive connect" denied_mask="send connect" addr=none peer_addr="@/tmp/.X11-unix/X1" peer="unconfined" [15:04] (once I comment out the X abstraction in the security policy) [15:04] attente, do you know how the gtk backend detection works? [15:05] what does it look for to know if it's under wayland? [15:05] attente: if you were so inclined, tossing a snap in --edge with wayland would be cool if you had time [15:05] jdstrand: sure, it does take a while to build though [15:05] jdstrand, ghex uses gtk from the archive which has the wayland backend included [15:05] attente, jdstrand, I think ghex is fine, it's just that gtk probably goes "neh, can't use wayland, let's use x11" [15:05] seb128: i think it just tries to connect to the wayland socket and if it fails, moves on to trying x11 [15:06] seb128: could be [15:07] seb128: why would it do that? there are no apparmor denials for anything other than X [15:07] weird [15:08] jdstrand, does it make a difference if you GDK_BACKEND=wayland ? [15:08] it's supposed to check wayland first [15:08] thinking that maybe it copied something into SNAP_USER_DATA, I deleted ~/snap and tried again [15:09] seb128: it just says: [15:09] Cannot open display: [15:09] attente, do you know where is the check code in gtk? [15:09] it looks like without GDK_BACKEND it tried mir first [15:10] seb128: https://git.gnome.org/browse/gtk+/tree/gdk/gdkdisplaymanager.c#n265 [15:10] it iterates through there [15:11] I don't know if anything is missing, but: http://paste.ubuntu.com/24328093/ [15:11] the actual checking for wayland is done here: https://git.gnome.org/browse/gtk+/tree/gdk/wayland/gdkdisplay-wayland.c#n512 [15:12] if wl_display_connect fails, it moves on to the next backend [15:12] jdstrand, is XDG_RUNTIME_DIR correctly set under snappy nowadays? [15:14] jdstrand, https://www.mankier.com/3/wl_display_connect [15:14] seb128: XDG_RUNTIME_DIR=/run/user/1000/snap.ghex-udt [15:14] it checks for the socket in XDG_RUNTIME_DIR [15:14] is there a wayland socket in that dir? [15:14] so, the socket isn't in there [15:14] k [15:14] there you go [15:14] dconf is in there [15:15] huh, that'll be an interesting problem [15:17] seb128: that's indeed it [15:17] apparmor="DENIED" operation="connect" profile="snap.ghex-udt.ghex" name="/run/user/1000/wayland-0" pid=16579 comm="ghex" requested_mask="wr" denied_mask="wr" fsuid=1000 ouid=1000 [15:17] attente: I don't need the edge snap [15:17] good [15:17] jdstrand, what did you change to get those DENIED? [15:17] great, it's kind of a pain to build :) [15:18] set the XDG_RUNTIME_DIR to the other location? [15:18] seb128: well, good, except that dsert advised us to change XDG_RUNTIME_DIR to /run/user/1000/$SNAP_NAME for dconf, but wayland isn't there [15:19] jdstrand, seems like we had hacks in the desktop launcher for mir, https://github.com/ubuntu/snapcraft-desktop-helpers/blob/master/common/desktop-exports#L43 [15:19] jdstrand, we might need something similar for wayland [15:19] seb128: install ghex, modify its profile to comment out the X abstraction include, load the profile, do 'snap run --shell ghex-udt.ghex', export XDG_RUNTIME_DIR=/run/user/`id -u`, then $SNAP/command-ghex.wrapper [15:20] though ideally things would work out of the box [15:20] easy peasy [15:20] ;) [15:20] seb128: perhaps there is another env var that can be set? [15:20] I'm googling for that [15:20] doesn't seem so [15:21] https://wayland.freedesktop.org/docs/html/ch04.html mentions WAYLAND_DISPLAY [15:21] WAYLAND_DISPLAY I think is just the basename [15:21] but that's to change the name of the socket [15:22] not the dir [15:22] right [15:22] why don't we use a common XDG_RUNTIME_DIR? [15:22] yeah. the man page is very explicit that it has to be in XDG_RUNTIME_DIR [15:22] attente, allison[m], ^ do you know what flatpak does with that? [15:23] sorry, i've no clue [15:23] well, we could symlink [15:23] snappy is doing the wrong thing here [15:24] it shouldn't change the value of XDG_RUNTIME_DIR. i've had to work around that in my dconf work. [15:24] yean, that works [15:24] they bind mount it I think [15:24] flatpak doesn't have the host's XDG_RUNTIME_DIR in its mount namespace [15:24] but it does have a bindmount of a subdirectory of it, at the same location in both [15:25] snap run --shell ghex-udt.ghex. 'cd $XDG_RUNTIME_DIR ; ln -s ../wayland-0 . ; $SNAP/command-ghex.wrapper' [15:25] so you'll see two different /run/user/1000/ directories on the inside and outside [15:25] apparmor="DENIED" operation="connect" profile="snap.ghex-udt.ghex" name="/run/user/1000/wayland-0" pid=16624 comm="ghex" requested_mask="wr" denied_mask="wr" fsuid=1000 ouid=1000 [15:25] but /run/user/1000/app/org.gnome.gedit/ will be the same [15:25] that makes sense to me [15:25] https://github.com/flatpak/flatpak/blob/master/common/flatpak-run.c#L1988 [15:25] allison[m]: not that snappy doing the wrong thing there came at dsert's explicit request [15:25] note* [15:26] i pushed the snappy team to do something similar and allow $XDG_RUNTIME_DIR/snap.snap-id to be permitted in a similar way [15:26] but they additionally reset the value of XDG_RUNTIME_DIR to this new directory -- messing everything up in the process [15:26] that second change should be backed out [15:26] jdstrand: no. i never asked for XDG_RUNTIME_DIR to be changed to a different value [15:26] i only asked for a new directory to be created and allowed through apparmor [15:27] seb128: I don't know if you saw in the backscroll, but the desktop part could just do a symlink from $XDG_RUNTIME_DIR/../wayland-0 to $XDG_RUNTIME_DIR/wayland-0 while the proper way this should be handled is worked out [15:28] jdstrand, no I didn't, thanks for rementioning it [15:28] i've made 3 comments in https://bugs.launchpad.net/snap-confine/+bug/1620442 and none of them say anything about changing the envvar [15:28] allison[m]: oh I didn't realize you changed your nick [15:28] Ubuntu bug 1620442 in Snappy "snap fails because XDG_RUNTIME_DIR is set to /run/user/1000" [High,Fix released] [15:28] * jdstrand gets the bug [15:29] maybe i made the mistake of commenting on a bug with a "we need this" when the original request was not exactly right [15:29] allison[m]: https://bugs.launchpad.net/snap-confine/+bug/1620442/comments/3 [15:29] Ubuntu bug 1620442 in Snappy "snap fails because XDG_RUNTIME_DIR is set to /run/user/1000" [High,Fix released] [15:29] allison[m]: that was in response (I thought) to my comment 2, which said to set XDG_RUNTIME_DIR [15:30] okay [15:30] anyway, this can be adjusted, just need to work through how [15:31] it does look like i ACK your proposal to do exactly that [15:31] so sorry for that [15:31] honestly, you need to do what flatpak does [15:32] ie: use mount namespaces to set up a clean copy of the xdg_runtime_dir at the same path, with the subdir bindmounted through [15:32] what does it do? a bind mount so the subdir of //run/user/uid/$SNAP_NAME is put on /run/user/uid? [15:32] please do not change the effective location of this directory on the inside/out [15:32] I just linked to the code [15:34] jdstrand: no. [15:34] /run/user/1000 ends up being a clean new tmpfs inside the snap [15:35] the subdir is shared, and it is at the same location inside and out [15:35] Laney, where? [15:35] where did I link it? [15:35] ups [15:35] Laney, unping, saw it :p [15:36] ok [15:36] anyway, desrt is explaining it with words [15:36] should i write this stuff down somewhere? [15:36] would be good I think [15:37] not sure what's the best somewhere though? [15:37] maybe a launchpad bug against snapd [15:37] * ogra_ points to forum.snapcraft.io [15:37] a README inside the source :) [15:37] allison[m]: I can do that. what I've just hear is not complete though (aiui). sorry if being dense. you said create a clean dir at /run/user/1000, with an empty subdir. that's fine, but wayland-0 isn't in there yet-- guess you left out that step? [15:38] jdstrand: bind mount it from the outside [15:38] jdstrand: bindmount it [15:38] ok, with the code Laney pointed at [15:38] jdstrand, did you see the github flatpak url from Laney some minutes ago? [15:38] right [15:38] :) [15:38] it wasn't clear that was the 3rd step [15:39] an alternative may be to make wayland add sockets per-client [15:39] we want to go this direction with dbus.... i'd like to chat about that with you guys, in fact [15:39] we had an interesting discussion at the gtk hackfest [15:41] it is certainly possible to do 3 steps that were outlined. it does mean that the setuid binary has to look for the wayland socket... if it isn't named wayland-0 (or more precisely, if there is no or more than one wayland-N, then have special casing [15:42] s/wayland-N,/wayland-N),/ [15:43] let me put this in the forum. seb128, would you be able to adjust the desktop part to do the symlink hack while the proper solution is designed and implemented? [15:45] jdstrand, sure, I'm about to step out for a bit but I can have a look at that later/tomorrow [15:46] seb128: shell snippet: cd $XDG_RUNTIME_DIR ; test -S $XDG_RUNTIME_DIR/../wayland-0 && ln -s $XDG_RUNTIME_DIR/../wayland-0 . [15:46] k, noted, thanks [15:46] seb128: that's fine, no rush, just curious on the status of things [15:47] seb128: note that the above is allowed by today's policy. actually using wayland-0 is not yet [15:56] \j [16:04] back in a couple of hours [16:04] off to do some zen digging [16:11] allison[m], attente, seb128, Laney: fyi, https://forum.snapcraft.io/t/wayland-dconf-and-xdg-runtime-dir/186 [17:51] night all [21:31] The best silver lining if this is the end of Mir is that people will stop incorrectly referring to it as MIR [21:42] * qengho works on the Tor people. [21:52] we have to get rid of SystemD next then [21:52] * Laney grimaces [22:20] Laney, YES... [23:09] andyrock: yo [23:10] a1fa: hey [23:10] you make it out alive or they still chopin' heads over there? [23:14] a1fa: still alive :D [23:44] great to hear [23:51] having to adjust to gnome a year in advance ;) is it blasphemy if your gnome shell looks like unity? [23:52] I'd argue that GNOME shell is extendable for that reason, so you can make it work however you want. [23:52] can you move the time over to the right? [23:53] so stupid to put it in the middle of the screen [23:54] There is probably an extension to do that, and if not, it would not be too hard to write one.