/srv/irclogs.ubuntu.com/2017/08/09/#ubuntu-desktop.txt

didrocksgood morning05:50
oSoMoNgood morning desktoppers!06:12
oSoMoNduflu, should we mention the test chromium packages (https://launchpad.net/~osomon/+archive/ubuntu/cr-vaapi-test/+packages) on http://wiki.ubuntu.com/IntelQuickSyncVideo ?06:14
=== JanC_ is now known as JanC
didrockshey oSoMoN06:19
oSoMoNsalut didrocks06:19
seb128good morning desktopers06:49
oSoMoNgood morning seb12806:49
seb128lut oSoMoN, comment ça va ?06:50
oSoMoNbien, et toi?06:50
oSoMoNla nuit a été plus longue qu’hier?06:50
seb128bien, pas encore réveillé par contre06:50
seb128un peu ouais, mais elle aurait pu être plus longue :p06:50
oSoMoNelles pourraient toujours être plus longues :)06:51
seb128:-)06:51
didrocksre seb12806:56
dufluoSoMoN, sorry you log on usually when I'm at lunch... Yes I would mention it on the wiki after confirming it works06:57
dufluAlso... good morning didrock, oSoMoN, seb12806:58
didrockshey duflu!06:58
seb128re didrocks06:58
seb128hey duflu, how are you?06:58
dufluseb128, today is a better day :) You?06:59
oSoMoNduflu, I successfully tested that it works on my machine yesterday (you’ll need to turn on the #enable-accelerated-video flag in chrome://flags, it's off by default)06:59
dufluoSoMoN, Thanks, I will look07:00
seb128duflu, I'm not fully awake and I've a bit too many things on my todolist but I'm fine otherwise07:02
seb128reading about meson build system07:02
seb128jbicha, those GNOME updates switched to meson are not translators friendly :-/07:02
dufluOh I forgot I'll need to refresh the Kaby Lake machine. That's the only one that does VP9, and VP9 seems to be the only codec YouTube likes to use these days07:04
dufluoSoMoN, does that include intel's proposed patch?07:13
dufluAh yes I see07:14
oSoMoNduflu, yes07:14
oSoMoNduflu, I haven't tested VP9, but http://distribution.bbb3d.renderfarming.net/video/mp4/bbb_sunflower_1080p_60fps_normal.mp4 was decoded and rendered fine at ~12% CPU on my X23007:15
dufluoSoMoN, yeah cool. That's the perfect test for Ivy Bridge07:15
dufluoSoMoN, tried a Wayland session? I'm seeing the usual libva fails to init in a Wayland session. This causes a long delay on startup, and probably means VAAPI won't work there (because Chrome is an X app and libva requires DRI2, not DRI3 in Xwayland)07:23
duflu(you can see the error if you run chromium-browser from a terminal)07:24
dufluoSoMoN, Yes it's only working in Xorg sessions. We'll need to get a Chromium fix in as part of bug 1698287. Sadly there is no workaround to make it work in a Wayland session because it's an X11-only app07:34
ubot5bug 1698287 in mpv (Ubuntu) "VA-API fails to initialize in a Gnome Shell Wayland session" [Medium,In progress] https://launchpad.net/bugs/169828707:34
dufluThe fix is to just tell the app to try the wayland platform before trying x1107:34
dufluIt's also a bit unstable but that's not surprising for Chromium 6207:38
oSoMoNduflu, yeah07:39
dufluoSoMoN, still looking good considering it's early days. That same initialization bug needs fixing in multiple locations07:40
dufluoSoMoN, actually a proper fix for Wayland sessions might not come till Chromium moves from X11 to Wayland. Do you know if people are working on that still?07:47
oSoMoNno idea, I'll need to check again07:47
dufluIt could also be fixed if someone enhanced libva to support DRI2, but that feels like investing in old technology07:48
dufluI mean to support DRI3. It already supports DRI207:49
dufluMaybe not so old07:49
oSoMoNyeah, I’ve seen in the bug report that you consider this unlikely to happen07:49
dufluoSoMoN, don't know. Seems there are multiple Intel engineers working on the various related projects07:50
dufluAnything could happen07:50
oSoMoNit'd be good to have a POC at intel to inform us about their plans07:51
dufluoSoMoN, anyway if you want to play local MP4 files I will soon have totem and mpv fixed for out-of-the-box support in both Xorg and Wayland07:56
dufluI mean I have the fixes. Just distro patches to do07:56
oSoMoNnice07:56
oSoMoNmorning willcooke07:57
oSoMoNwillcooke, https://forum.snapcraft.io/t/sharing-files-via-tmp/1613 to follow up on our conversation yesterday07:57
willcookeo/08:01
oSoMoNhttps://blogs.igalia.com/tonikitoo/2017/05/17/chromium-musozone-update-h12017-wayland-x11/08:05
willcookeafternoon duflu.  I'm getting this error: http://paste.ubuntu.com/25270562/08:05
willcookewait, hold on - I think I was in a Wayland session as well, even though I told it not to08:06
dufluwillcooke, good morning. You just missed me discussing that error with oSoMoN. It is https://bugs.launchpad.net/ubuntu/+source/intel-vaapi-driver/+bug/169828708:07
ubot5Ubuntu bug 1698287 in mpv (Ubuntu) "VA-API fails to initialize in a Gnome Shell Wayland session" [Medium,In progress]08:07
dufluBut there is no simple workaround for Chromium :/08:07
dufluOther than "Log in to X"08:07
seb128hey willcooke08:08
willcookeCPU usage goes from ~ 100% to ~ 50%08:11
willcookeso that's a win08:12
dufluwillcooke, oh good news in that area: https://www.igalia.com/nc/igalia-247/news/item/three-new-people-join-igalia/08:12
dufluEleni was working on Chromium with Andy for a while08:12
dufluseb128, is it better if my future sponsorships are not assigned? Do people get confused if they are assigned?08:17
oSoMoNduflu, from that igalia reported dated from the 17th of May, there's active development and progress, but it doesn't look like it's been upstreamed yet08:19
willcookehttp://imgur.com/a/dG08X08:19
willcookebefore on the left, after on the right08:19
dufluoSoMoN, It is a hard problem. Chromium basically needs to become a display server, and each tab is a client08:21
dufluwillcooke, I was hoping for a bit lower. What CPU is that?08:21
dufluAh, never mind. I get about the same (adjusted for the fact I'm looking at a desktop server CPU)08:25
seb128duflu, no, I think assigned is fine, did anyone comment saying it's confusing?08:29
seb128duflu, sponsors tend to assume that if it's assigned to somebody who has upload right then that person is going to deal with the upload, but since you don't have upload rights there should be no confusion08:29
dufluseb128, I just recall seeing something on that topic in related wiki pages (syncs and SRUs).08:30
dufluAlthough these are not syncs or SRUs08:30
seb128right08:30
seb128well usual rule is to not assign to the people you need08:30
seb128e.g don't assign the bug to sponsors or sru team08:31
seb128just subscribe08:31
seb128but keeping it assigned to yourself is fine08:31
willcookemorning08:44
willcookesomething odd happenened just now08:44
willcookedid you guys see my messages from 09:05 and 09:19?08:45
willcookeit's now 09:4508:45
seb128willcooke, I saw08:46
seb128<willcooke> CPU usage goes from ~ 100% to ~ 50%08:46
seb128<willcooke> so that's a win08:46
seb128<willcooke> http://imgur.com/a/dG08X08:46
seb128<willcooke> before on the left, after on the right08:46
seb128hum, python packaging question08:48
seb128if a package is shipping python files to a private dir, is it useful to byte compile those and if so is that something the tools can do for you or that you need a postinst/rm for?08:49
seb128e.g https://github.com/fossfreedom/alternative-toolbar/blob/debian/debian/postinst08:50
seb128(cyphermox raised that one as a weird thing to do in his MIR review)08:50
seb128wb willcooke08:51
willcooke:)08:51
willcookestill better than slack08:51
willcooke;D08:51
dufluwillcooke, Yes we saw and replied. That's bug 169828708:56
ubot5bug 1698287 in mpv (Ubuntu) "VA-API fails to initialize in a Gnome Shell Wayland session" [Medium,In progress] https://launchpad.net/bugs/169828708:56
willcookeduflu, cool!  (Didnt see  replies cos netsplit)08:57
willcookeduflu, I've got to jump on another call now, will be late for bluetooth meeting (or might not make it, sorry) (cc seb128)08:58
dufluwillcooke, Not cool though. It requires a native port of the browser to Wayland really (if in a Wayland session)08:58
dufluOK, late is cool08:58
willcooke@ browser - Ah, I see08:58
dufluOr someone to greatly enhance libva08:59
dufluor Xwayland08:59
willcookeheh08:59
seb128duflu, chromium refuses to start for me so I'm going to be a bit late for the bluetooth meeting as well09:04
dufluNo one there but me09:04
seb128is Konrad not around this week?09:06
dufluseb128, well he's not in IRC or the hangout. But is writing emails09:10
dufluseb128, that's OK. Let me give you another debdiff :)09:10
seb128duflu, if neither Konrad or Will are around and I'm fighting with chromium (and firefox doesn't do hangout) let's skip09:10
seb128haha09:10
dufluYou think I'm kidding09:11
* seb128 hides09:11
dufluCrap. I closed the hangout by mistake09:14
dufluseb128, no hurry. After the meetings I'm at EOD... https://bugs.launchpad.net/ubuntu/+source/mpv/+bug/170810209:16
ubot5Ubuntu bug 1708102 in mpv (Ubuntu) "Enable hardware video decoding by default in mpv" [High,In progress]09:16
seb128duflu, let's skip the hangout as said, unless willcooke finishes his other meeting and want to chat a bit there09:16
dufluThat's fine. I'm still there in case Will has time09:17
seb128duflu, debdiff looks fine, it's usually better to list the files you change in the debian/changelog though09:18
seb128like instead of "* Enable hardware acceleration by default"09:18
seb128you would have09:18
seb128* debian/mpv.conf.ubuntu, debian/rules:09:18
seb128  - Enable hardware acceleration by default09:18
dufluseb128, I was aiming to make readable changelogs. I thought that style was low value but I can do both in future09:18
dufluToo many people write unreadable changelogs09:19
seb128fair09:19
seb128the details on what files are impacted by a change make merges on debian a bit easier though09:19
dufluYeah09:19
seb128especially when there is no packaging vcs to dig the commits09:19
dufluseb128, I can make another one if we're not meeting09:20
seb128duflu, don't bother redoing that one09:20
seb128it's for next time09:20
seb128unping about my python packaging question from earlier if anyone read that09:36
dufluI'm sure that's not a verb09:38
dufluNeither is ping though09:38
pittiit so is a verb amongst IRC people :)09:39
seb128hey pitti! wie gets?09:39
pittiit's even a (made-up) strong one :) ("he pung", as some people like to say ☺ )09:39
pittiseb128: ça va seb128 ! merci, je vais bien !09:39
seb128ça va bien ici aussi merci !09:40
pittiseb128: j'attends avec impatience nos vacances, ils commencent ce vendredi pour trois semaines09:40
dufluHi pitti09:44
pittiduflu: hello! *wave*09:44
seb128pitti, où est-ce que vous allez en vacances ?10:19
pittiseb128: on va à la côte de Ouest des État-Unis10:21
pittiseb128: for the total eclipse, pour des parcs nationals, pour visiter l'usine de Boeing, etc.10:21
seb128ah, nice10:22
seb128pitti, enjoy!10:22
pittiseb128: merci ! je suppose que vous reste chez vous, avec votre petit enfant ?10:23
seb128pitti, oui, ici et visiter nos familles en France10:24
seb128popey, hey, change the dbus monitor command from yesterday by "dbus-monitor interface='org.freedesktop.Notifications'"11:13
jbichaseb128: is it meson or is it LP: #1688994 ?11:20
ubot5Launchpad bug 1688994 in pkgbinarymangler (Ubuntu) "dh-translations: cannot build POT without intltool" [High,New] https://launchpad.net/bugs/168899411:20
jbichagood morning11:20
seb128jbicha, that's the same thing11:22
seb128is that there seems there is no equivalent to "intltool-update -p" in the new build systems11:22
seb128I can't find how to update it manually with one command equivalent to ^ either11:23
seb128then if there is one we need to update the tools11:23
jbichaintltool was already being dropped by GNOME before the meson switch :(11:23
jbichabut I think meson requires dropping intltool? so it's getting worse :(11:24
seb128right, well under autotools&gettext we had standard make targets still11:25
seb128that seems to not even be the case anymore with meson11:26
seb128I asked about it on the gnome #i18n channel, let's see11:26
seb128otherwise I guess we are going to need to overwrite the dh_translations call in debian/rules for a bunch of packages until there is a proper solution11:26
seb128that wouldn't be the end of the world, that set of package is limited11:26
seb128still it's more work and annoying11:27
jbichalet me know when you have the debian/rules override and I can start adding it to packages11:28
seb128will do, thanks11:29
jbichaI'd like to cherry-pick https://github.com/hughsie/appstream-glib/commit/9df56f26f11:29
seb128one for Laney11:29
jbichait might cause a few packages to ftbfs because the dh_install rules assume the old directory but I'd rather fix that now than in a few weeks11:30
seb128right11:30
seb128popey, do you use "solaar"?11:31
popeyi don't know what that is11:32
seb128so no11:32
seb128"Solaar is a Linux device manager for Logitech's Unifying Receiver peripherals"11:32
seb128upstream guessed it might lead to issue like your power notification11:32
popeyno then, i have no logitech devices11:32
seb128popey, anyway, did you see my updated command? please get the log and then we hopefully know what the issue is :-)11:32
popeyah, you updated it? let me see11:33
popeyi have been draining the laptop battery for a bit so I can plug it in and capture any output, will update the command to make sure I get the right stuffs11:33
seb128thanks11:34
jbichaseb128: should I stop doing 3.26 updates until the translation problem is fixed/worked around?11:34
seb128jbicha, no, keep going, we still have the current template so worth cases we miss a few strings and I'm going to sort that out in the coming week11:34
jbichaok11:35
flocculanttkamppeter: seeing bug 1709572 on Xubuntu since your change last Friday11:35
ubot5bug 1709572 in system-config-printer (Ubuntu) "system-config-printer.py crashed with ValueError in require_version(): Namespace Secret not available" [Medium,Confirmed] https://launchpad.net/bugs/170957211:35
jbichajibel: could you consider adding installing Chrome to the gnome-software test plan? see LP: #1708936 or LP: #167242411:36
ubot5Launchpad bug 1708936 in gnome-software (Ubuntu) "GNOME Software 3.20.5 doesn't install 3rd party deb's, doesn't prompt for authentication" [Critical,Confirmed] https://launchpad.net/bugs/170893611:36
ubot5Launchpad bug 1672424 in gnome-software (Ubuntu Zesty) "Cannot install Debian files outside of the repositories" [Critical,Fix released] https://launchpad.net/bugs/167242411:36
seb128flocculant, it should have a depends added on gir1.2-secret-111:37
seb128tkamppeter, ^11:37
seb128on that note it's lunch time, bbl11:37
flocculantseb128: thanks - adding that package and all is well with the world :)11:38
popeyseb128: success! :D12:01
popey(updated bug 1709194)12:01
ubot5bug 1709194 in gnome-power-manager (Ubuntu) "Power notification appears with no detail" [High,Incomplete] https://launchpad.net/bugs/170919412:01
popeywillcooke: that issue I mentioned about apps taking 10+ seconds to start. found an AU post which has a solution. kill all apps, kill gnome-keyring-daemon, re-launch and then launch apps, everything launches instantly!12:15
popeyhttps://askubuntu.com/questions/788075/ubuntu-16-04-some-applications-take-too-long-to-start-up12:15
popeyworth remembering!12:15
jbichapopey: do you have a bug # for that?12:18
popeynot yet, but it only happened today and I'm on 16.0412:19
popeywill file it in moment12:19
jbichadoes that issue happen with 16.04 too?12:20
popeyI am on 16.04, and it happened here. willcooke said he'd seen it on 17.xx at some point too. I have only seen it on 16.0412:21
popeybecause I am a heavy 16.04 user, and only lightly use 17.xx12:21
popeybug 1689825 looks to cover it12:27
ubot5bug 1689825 in lightdm (Ubuntu) "gnome-keyring not unlocked on boot" [Undecided,Confirmed] https://launchpad.net/bugs/168982512:27
popeypeople suggesting removing/purging dbus-user-session as a workaround/solution12:27
pittipopey: dbus-user-session will totally break 16.0412:29
pittiwe only started work on that for 16.1012:29
jbichapopey: the trigger for that bug is the flatpak ppa, I emailed Alexander Larsson about it a few weeks ago and got no response yet12:29
popeyoof12:30
popeyi do have that ppa on my system, good catch!12:30
popeydo we need an upstream bug?12:31
jbichaupstream is talk to the PPA dev, maybe you'll have better luck than I did ;)12:33
popeyheh, okay12:34
popeychallenge accepted12:35
jbichapopey: do you have a Facebook account? would you be interested in verifying this bug https://bugzilla.gnome.org/show_bug.cgi?id=785726 and its SRUs?12:37
ubot5Gnome bug 785726 in general "facebook: Cannot add a new account with Graph API > 2.3" [Normal,Resolved: fixed]12:37
jbichausing GNOME Online Accounts12:38
popeyjbicha: will take a look12:38
jbichaI didn't prepare the SRUs yet12:39
popeyjbicha: lemme know when you have a deb I can try?12:39
jbichacould you verify the bug first? ;)12:39
jbichaI don't have a FB account12:39
popeyoh, hah, on 16.04 or 17.10?12:41
jbichaeither one, I haven't fixed it on 17.10 yet because I wanted to verify the bug before fixing it12:42
popeyok, will try both12:43
popeyjbicha: success. https://twitter.com/gnomealex/status/89526369098292838412:49
popeyjbicha: that facebook bug isn't exactly clear from a user perspective what the effect is. I connected my 17.10 system to facebook. Not sure how to confim whether it did or not.12:54
jbichaif you can still connect your account, then it might not be an important enough bug to SRU12:56
popeyit certainly seems a bit broken. Now I added an account I can't view it or add another12:58
jbichapopey: could you file a LP bug for the FB issue that I can use for SRU tracking?13:01
tkamppeterflocculant, seb128, bug 1709572 fixed.13:04
ubot5bug 1709572 in system-config-printer (Ubuntu) "system-config-printer.py crashed with ValueError in require_version(): Namespace Secret not available" [High,Fix committed] https://launchpad.net/bugs/170957213:04
jdstrandkenvandine (cc willcooke and jamesh): fyi, https://github.com/ubuntu/snapcraft-desktop-helpers/pull/68. I'm not sure who is tasked with getting snaps to run under wayland, but that PR gives some clues13:05
jdstrandkenvandine (cc willcooke and jamesh): you might want to see https://forum.snapcraft.io/t/wayland-dconf-and-xdg-runtime-dir/186/1013:06
kenvandinejdstrand, thx!13:06
kenvandinenot sure who is working on that, but i'll make sure this gets looked at13:06
jdstrandkenvandine (cc willcooke and jamesh): note that PR shouldn't be committed as is (see comment I added today), but it highlights the things to be considered. the env vars and the symlink (we may not always need the symlink, see later responses to the forum post)13:07
jdstrandkenvandine (cc willcooke and jamesh): I didn't look very hard, but it seems that gdk-pixbuf-query-loaders may be sensitive to the wayland env vars13:08
popeyjbicha: bug 170962113:08
ubot5bug 1709621 in gnome-online-accounts (Ubuntu) "Facebook login appears broken" [Undecided,New] https://launchpad.net/bugs/170962113:08
jdstrandkenvandine (cc willcooke and jamesh): not saying anything to be done there-- just keep an eye out for it13:09
kenvandinedidrocks, ^^13:10
kenvandinedidrocks, is that desktop helpers PR for wayland support something you can look at?13:10
jdstrandthanks for helping coordinate that work13:11
kenvandinejdstrand, thanks for pointing it out13:11
didrockskenvandine: not nowdays, I'm on default experience for our desktop13:12
didrockskenvandine: I think you are way closer than I am now due to handling the platform snap, are you happy to look at this?13:12
kenvandinesure13:12
jdstrandkenvandine: since you are looking at it, note that jamesh and I have been discussing how this might fit into his portals work which would remove the need to setup the symlink, but that is farther out13:16
jdstrandkenvandine: that's all in the forum13:17
kenvandinejdstrand, thx13:17
kenvandinejdstrand, yeah doesn't help for 17.1013:17
jdstrandkenvandine: I submitted https://github.com/snapcore/snapd/pull/3690 yesterday (for wayland interface). I expect that to land quickly13:17
jdstrandkenvandine: well, afaik his portals work is 17.10 material, but its snapd, so it can float in at any time13:18
jdstrandkenvandine: when pr 3690 lands, it'll need something along the lines of the snapcraft-desktop-helpers pr for people to use it13:19
jdstrandkenvandine: lastly, I am creating preliminary new interfaces for gnome-shell that uses only modern apis (tentatively called 'desktop')13:20
kenvandinejdstrand, so wayland as a new interface means app developers will need to specify that interface in their snap.  I fear some apps that should work will just fail13:20
jdstrandkenvandine: eg, no dbusmenu, etc, etc13:20
kenvandinewouldn't it be better for the desktop interface to support both wayland and x11?13:20
kenvandineie app developers not care if it's wayland or x11, it's just a desktop13:21
jdstrandkenvandine: this is a complicated topic but is following the existing paradigm13:21
jdstrandkenvandine: that is true, but we want to be able to have people disconnect them13:21
jdstrandie, if I'm using wayland, I should be able to disconnect the x11 interface13:22
kenvandineyeah, maybe super-interfaces?  that bundle them13:22
kenvandinebut i bet this has been discussed :)13:22
jdstrandthat is sorta what unity7 is now, and I think tht is a mistake13:22
jdstrandunity7 includes access for x1113:23
jdstrandin that case, it made some sense-- it couldn't run without it13:23
jdstrandbut with the new 'desktop' interfacec, it could run with either wayland or x1113:23
kenvandinefrom a desktop users perspective, they usually aren't going to understand the difference in x11 or wayland or even know which they are running13:24
jdstrandkenvandine: my plan is to get the policy together, submit the PR and then ping you and jamesh to participate in the discussion13:24
jdstrandkenvandine: that's fine-- the developer will likely do [plugs: wayland, x11, desktop]13:24
kenvandinejdstrand, sounds good, thanks13:24
jdstrandkenvandine: they'll all autoconnect, but then a security conscious person can 'snap disconnect foo:x11'13:25
kenvandineok13:25
kenvandinethat's reasonable13:25
jdstrandI think it is reasonable for a developer to specify where it is expected to work13:26
oSoMoNchrisccoulson, when opening an attachment in thunderbird or a file in firefox, the file is written to /tmp and handed over to xdg-open (I assume). Do you know whether we can easily change the default location so that it would be somewhere under $HOME ?13:26
jdstrandsince they have to test in all those places anyway13:26
jdstrandin the case of gnome-shell, it probably makes sense to plugs x11 by default in this transition period because mutter depends on Xwayland for a few things and wayland snaps sometimes connect to Xwayland for things13:28
seb128tkamppeter, thanks for fixing the s-c-p missing depends13:29
jdstrandtrying to force not using x11 would be a poor user experience. but, to support moving forward, we should be able to let people disconnect x11 so people can try it out13:29
jdstrandkenvandine: ^13:29
kenvandinejdstrand, you might want to leave appindicator access13:29
kenvandineyeah13:29
seb128jdstrand, you were working on snaps under wayland so we stayed out of it ... should we pick it up?13:29
seb128kenvandine, ^13:29
kenvandineseb128, jdstrand did the interface13:29
kenvandinei need to tweak some stuff for desktop helpers13:29
kenvandineseb128, i'm going to pick up that part13:30
flocculanttkamppeter: thanks :)13:30
kenvandinebut i guess i need the interface to land13:30
kenvandineor learn how to run my own snapd again :)13:30
kenvandinei had a script for that at one point :)13:30
jdstrandseb128: I've only been working on a part of it. willcooke and I sync'd at the product sprint. I'm tasked with the wayland interface and a base set of policy for the new 'desktop' interface. getting the snapcraft-desktop-helpers to work, the gnome platform snap, etc, etc is for the desktop team13:30
kenvandineseb128, and jdstrand had some good pointers for what needs tweaking13:31
jdstrandI really needed to get something going though, so I did that snapcraft-desktop-helpers PR to help you13:31
seb128kenvandine, ok13:31
jdstrand(I was blocked)13:31
kenvandinejdstrand, we appreciate it :)13:31
seb128jdstrand, right, fair enough, I though the wayland interface didn't land yet and was a pre-requirement to test the other bits13:32
kenvandinejdstrand, i'm having a terrible time with classic snaps... i had 2 classic snaps yesterday that ran one time13:32
kenvandinebut won't start again :/13:32
jdstrandseb128: it hasn't landed, but is in a PR as of yesterday. it willland soon13:32
kenvandineseb128, it's in a PR13:32
jdstrandit is currently a single rule, so it is easy to test. snap install foo; snap connect foo:gnome-3-24-platform ... ; edit /var/lib/snapd/apparmor/profiles/snap.foo.foo to have the wayland rule ; apparmor_parser -r ... ; launch the app13:33
jdstrandI'm happy to help with advising how to test stuff13:34
kenvandinejdstrand, ok13:34
jdstrandkenvandine: oh, one other thing13:34
jdstrandkenvandine: running gnome-sudoku under wayland on xenial results in: (gnome-sudoku:3064): Gdk-WARNING **: Wayland compositor does not support xdg_shell interface, not using Wayland display13:35
kenvandinejdstrand, but that's after i'm running your snapd branch right?13:35
jdstrandkenvandine: if you use my snapd branch you don't have to do anything special (it adds the rule)13:36
kenvandineoh, i can add the rule without rebuilding snapd?13:36
jdstrandkenvandine: oh, well, you need to plugs wayland :)13:36
kenvandineah, just apparmor :)13:36
jdstrandkenvandine: yes. if you don't want to rebuild snapd, just tack the rule on the profile13:36
kenvandineah, cool!13:36
jdstrandkenvandine: snapd will refresh the profile though on reboot, reinstall, connect/disconnect, etc, so be aware of that13:37
ogra_cheating !13:37
ogra_:)13:37
kenvandinethat's a great cheat though13:37
jdstrandit's convenient13:37
jdstrandI use it when developing interfaces to quickly iterate13:38
jdstrandwhen I'm happy, I copy it all over to the interface and then build a new snapd, retest13:38
jdstrandkenvandine: you saw the gnome-sudoku comment with wayland on xenial comment above? ^13:39
kenvandineyeah, i'll need to look into that13:39
kenvandinei guess it fell back to xwayland?13:39
jdstrandseb128, kenvandine: this actually worries me since I think it gets into the conversation with niemeyer re the gnome-3-24 snap. using gnome-3-24 with wayland doesn't work with xenial's wayland13:40
jdstrandkenvandine: well, no, cause I removed rules allowing to talk to X :)13:40
kenvandine:)13:40
kenvandinejdstrand, well we won't be running wayland on xenial13:40
jdstrandkenvandine: in *my* testing, I didn't plug unity7 or x1113:41
jdstrandkenvandine: then the desktop helper will likely need to be even smarter13:41
jdstrandkenvandine: because there are only series 16 snaps today13:42
kenvandineindeed13:42
kenvandinebut 17.10 will be wayland and 16.04 will be x1113:42
jdstrandfor Ubuntu default install13:42
jdstrandsnaps go way beyond that13:42
kenvandineyes13:42
kenvandinebut for the average desktop user, 16.04 users will be running unity713:43
jbichapopey: shotwell doesn't support GOA yet13:43
jdstrandI agree it is unlikely to have many users on 16.04 using wayland in Ubuntu. but, derivatives or other distros that have an incompatible wayland with what is in the gnome-3-24 platform snap-- these are considerations13:43
jdstrandI wonder if there is a way to query the wayland socket for capabilities or version or something13:44
jdstrandI don't know if mutter supports that13:45
jdstrandbut it is a little concerning that we already see an incompatibility between what is in 16.04 and what is in gnome-3-2413:46
kenvandinejdstrand, so you were running gnome-shell on 16.04 with wayland right?13:46
jdstrandkenvandine: yes13:46
jdstrandkenvandine: I abandoned the gnome-sudoku snap for testing that. I grabbed the gnome-logs-udt snap from the store (3.18 libs embedded) and hacked its wrapper to force wayland to test13:47
jdstrandthe env var setup and symlink from the pr I pointed you at all worked fine for that13:48
jdstrandbtw, there is a gnome-3-18 content snap in the store, but I don't know of anything that uses it13:48
kenvandinejdstrand, yeah13:49
jdstrandkenvandine: so, I think that's my full brain dump13:49
kenvandinejdstrand, appreciated!13:49
kenvandinei'll try to dig into that more13:50
jdstrandkenvandine: thanks!13:50
jdstrandkenvandine: oh, one more thing-- I wasn't able to locally snapcraft cleanbuild gedit or gnome-sudoku. probably because I was using xenial apt sources14:02
kenvandinejdstrand, yeah... you have to include the backport PPA14:03
kenvandinewhich i don't think you can do in cleanbuild14:03
kenvandineLP lets you build with the PPA though14:03
seb128shrug14:10
seb128jdstrand, those problems you listed on the forum are not specific to the platform are they?14:10
seb128jdstrand, we would have the same wayland on xenial issue with gtk bundled with the app14:11
seb128jdstrand, same for using dbus services14:11
kenvandineseb128, i think you are right14:12
jdstrandseb128: they are not specific to the platform in that any content snap might have related issues. why I believe it is relevant is that it doesn't work on the same series the base snap is even built from (ie, 16.04 with series 16 snapd). it then may be 'specialized' and something the publisher needs to make sure consumers use correctly14:13
seb128jdstrand, I don't understand how that has to do with "content"14:14
kenvandinejdstrand, the alternative for an app developer would be to build gtk and all the depends into their snap, which would run into the same issue14:14
seb128jdstrand, if you built gedit with that ppa bundling the libs you would have the exact same issues no?.14:14
jdstrandput another way-- you want the world to auto-connect to this content interface, but the world won't work properly14:14
seb128how so?14:14
seb128it would work as properly as if they bundled the same content14:15
didrocksit's the base issue that any flatpak or snap have, backward compatibility and services14:15
jdstrandthis conversation should be done in the forum. niemeyer needs to be present14:15
seb128jdstrand, the topic you commented on is that conversation14:15
jdstrandto me, a content snap that anything is allowed to connect to should reasonably work on the series it targets14:15
seb128but I don't understand what he tries to say14:16
jdstrandseb128: yes, but we are now talking here :)14:16
seb128and it's going in loop14:16
jdstrandniemeyer isn't here14:16
kenvandineit would work in the desktop environment provided in the LTS, gnome-shell is universe in 16.0414:16
jdstrandkenvandine: but snapd != Ubuntu14:16
seb128jdstrand, right, I though you might be able to make me understand what he means because the discussion is going nowhere on the forum14:16
seb128I guess I just going to give up14:16
jdstrandseb128: I am not representing niemeyer. I mentioned a concern I have about socket protocols and services14:16
seb128jdstrand, right, and I was asking why you think your concern about socket protocols and services wouldn't impact apps that bundle the same content14:17
seb128which is what I don't understand14:17
jdstrandbecause I don't think that the content interface design is handling well14:17
seb128there is no reason having the content mounted or bundled should make a difference14:17
seb128but what does it has to do with the content?14:18
kenvandineif gedit built all the reverse depends in the snap would have the same issue14:18
jdstrandthe content interface is meant to allow a specific publisher to bundle there stuff so that leaf snaps can consume it14:18
seb128the issue is that gnome 3.24 is made to work with recent wayland14:18
seb128if you were to bundle gnome 3.24 you would have the same issue14:18
jdstrandyou want to make a content interfacec for the world to use14:18
didrockssame with every services, and a shim if services aren't around14:18
didrocksthat's going to take some years though, from calendar to contact info to…14:19
jdstrandseb128: but if you bundle gnome3.24 yourself, it won't work on the series you are targeting14:19
jdstrandyu would need to change your snap to work14:19
seb128why not?14:20
jdstrandseb128: you just said it wouldn't14:20
jdstrand09:18 < seb128> if you were to bundle gnome 3.24 you would have the same issue14:20
seb128I don't understand how it's different from the content interface case14:20
jdstrandthe issue is that it doesn't work on wayland in 16.0414:20
seb128right14:20
seb128well that's a bug14:20
jdstrandso that is a bug in your snap14:20
kenvandineyeah, so any app using the newer gnome libs will have that problem14:20
jdstrandthat you would address14:20
jdstrandin your snap14:20
kenvandinenothing to do with the content interfae14:20
seb128the gnome-3-24 snap is targetting xenial and we should fix it to work on xenial14:20
seb128you found a bug14:21
ogra_all snaps are targeting xenial14:21
seb128but gedit with gtk built from upstream source would have the exact same issue14:21
ogra_and xenial only14:21
seb128right14:21
ogra_there is nothing else14:21
seb128exactly14:21
jdstrandif you have a content interface that claims to work everywhere (which auto-connecting strongly implies), then you need to make it work everywhere14:21
seb128so a snap bundling gtk would have the same issue14:21
* mdeslaur gets popcorn14:21
seb128it works on the serie you target14:21
kenvandineso i guess we need to be able to build the gnome libs with an older version of libwayland14:21
ogra_a snap bundling the very latest gtk would still be built against xenial14:21
seb128= xenial14:21
seb128ogra_, as does our platform snap14:21
ogra_right14:21
seb128so it's no difference14:22
ogra_thats the guaranteee that snaps give you14:22
jdstrandbut you've already said it isn't supported in certain situations. which is fine to say for your published snaps, but not necessarily ok for everyone else14:22
seb128the issue is that gtk 3.22 doesn't know how to talk to old wayland14:22
seb128jdstrand, I didn't say that14:22
jdstrand(I mean, maybe it is-- this is a conversation about how the content interface should work)14:22
seb128and I think it was a mistake for kenvandine  to say that14:22
kenvandineindeed ;)14:22
seb128gtk on xenial should be able to talk to the wayland from xenial14:22
jdstrandseb128: kenvandine said that 16.04's wayland is not supported14:22
jdstrandok14:22
jdstrandwell14:22
seb128that was an error to say imho14:22
jdstrandthis conversation should be in the forum14:23
seb128I'm not discussing more in the forum14:23
ogra_but then you cant hide it from niemeyer :P14:23
jdstrandI think it is important to understand whee niemeyer is coming from14:23
seb128it's like playing chess with a pigeon14:23
ogra_just speed up your pidgeons then :)14:23
ogra_jet-pidgeons FTW14:24
jdstrandit is simply that the content interface is designed for a particular use case-- sharing content for the same publisher. the publisher can dictate any terms she wants for her own snaps14:24
seb128jdstrand, I'm trying to understand but nobody is wanting to explain what they have in their head14:24
seb128jdstrand, it's fine and I'm good with that14:24
jdstrandnow, extending the content interface to the world has not been given a lot of though. so niemeyer is trying to work through that14:24
jdstrandthought*14:24
seb128then the snap team shouldn't have recommended us to build a platform using the content interface to reduce snap sizes14:24
jdstrandit isn't adversarial, it is trying to work through what everything means and to make sure everything works well going forward14:24
jdstrandwhy not?14:25
seb128because you just said it was not designed for that case14:25
jdstrandif Canonical is going to upload 20 gnome snaps then the content interface is perfect to support those 2014:25
seb128well it's not14:25
jdstrandanother publisher could do the same with their content snap and 250 snaps14:25
seb128apparently there is a problem with the "underline layer"14:25
jdstrandor whatever14:25
seb128which nobody would explain to us14:25
jdstrandit is when a particular snap is for everyone that things need to be thought through more14:26
seb128well if it's for our 30 apps then we might hit that underlining layer issue14:26
seb128so I would still like people to explain what the issue is14:26
kenvandinewe just want to make it as easy as possible for gnome/gtk app developers to snap their apps14:26
jdstrandseb128: yes, but it is limited to your snaps-- you can gate and fix and use edge and all those tools14:26
seb128there is no reason that a desgin problem would be an issue for external snaps and not ours14:26
seb128I don't want to break&fix14:27
jdstrandnone of those tools exist in ways tht work well when not from the same publisher14:27
seb128I want to understand the problem and don't break14:27
jdstrandof course14:27
seb128so please somebody tell us what that underlining layer issue is?14:27
jdstrandand niemeyer is trying to understand the issues to not break too14:27
seb128and what is the underlining layer?14:27
jdstrandI can't speak for niemeyer. please take to the forum14:28
seb128I did14:28
seb128a said that's going nowhere14:28
seb128so I guess it's end of discussion14:28
kenvandinei had said we would support this the same way we would support SRUs14:28
kenvandinesame level of support14:28
kenvandineso committed to fixing issues14:28
jdstrandlike I said-- I understand the angle he is coming from. I had my own separate concern. I recognize there are issues with the content snap design so participated in the conversation14:28
seb128there is no point discussing with somebody who doesn't want to discuss but just make his point14:28
seb128and be right14:28
seb128and ignore what you say anyway14:28
seb128jdstrand, -content14:29
seb128jdstrand, there are issues with the snap design14:29
seb128you have the same issues with the core snap and glibc14:29
jdstrandI'm sorry, but this is getting unproductive. I cannot represent him. I can't mediate the conversation when he isn't even present14:29
seb128nothing there is specific to the platform14:29
jdstrandif you are frustrated I suggest taking it to hangout14:30
kenvandineseb128, so back to my comment... do we want to support the gnome-shell/wayland scenario for 16.04?  I was really thinking our priority was what was in main14:32
kenvandinei hadn't been testing gnome-shell/wayland from 16.0414:33
kenvandinewell, the wayland interface wasn't available :)14:33
kenvandinebut i wasn't testing gnome-shell14:33
seb128I don't think wayland is usuable in 16.0414:35
seb128maybe we can build the gtk backport without that backend14:35
kenvandinethat's a good idea14:35
seb128could somebody try to install gnome-characters from gnome-software in artful and tell me if they also get a dialog asking them to enable universe?15:20
kenvandineseb128, sure15:21
seb128kenvandine, thanks15:21
kenvandineseb128, oh, i guess i should disable universe first?15:22
seb128no15:22
seb128that's my issue15:22
seb128I can apt install that deb and I've universe enabled15:22
didrocks(let me double confirm)15:22
seb128still it tells me that15:22
didrocksoh "starting"15:22
kenvandineseb128, didn't prompt me15:22
seb128k, thanks for testing15:22
seb128I can install other things from universe, it's weird15:23
didrocksdidn't prompt me either15:23
seb128thanks15:23
didrocksyw15:24
didrocksI have universe enabled on the same line in sources.list15:24
didrocksis it the same for you or separate lines?15:25
didrockslike "main universe…" all on a single line15:25
kenvandineseparate line here15:25
seb128separate15:27
didrocksso not that :/15:29
seb128https://launchpadlibrarian.net/332594099/universe.png15:30
seb128https://launchpadlibrarian.net/332594162/universe.png rather15:31
seb128well it's the same but I posted a duplicate so deleted one15:31
jbichaI believe seb has the g-software from artful-proposed installed, right?15:34
seb128sometime "installing" doesn't change15:34
seb128oh right, sorry15:34
seb128I forget that still in there only15:34
seb128https://bug786058.bugzilla-attachments.gnome.org/attachment.cgi?id=35727915:34
seb128jbicha, you do as well right? ;-) do you have the issue?15:45
jbichayes, no :)15:55
jbichaseb128: gnome-control-center just switched to the new "shell" in preparation for 3.25.9017:43
willcookenight all18:17
seb128jbicha, is it feature complete? from the discussions I had a GUADEC they didn't believe it had lot of testing yet21:43
seb128seems another low quality move from GNOME :(21:43
preppkk21:52
jbichaseb128: yes, it's feature complete, it just needs more testing and polishing22:32
Guest46914libreoffice 5.4 has been in debian unstable for 12 days, and debian import freeze isn't until 2017-08-24, so will it be imported for 17.10?22:32
jbichaGuest46914: yes, LibreOffice's release cycle is basically designed *for* Ubuntu's release cycle22:33
Guest46914oh wow, great! thanks!22:38

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!