[01:28] robert_ancell: I'm happy to look at any SRUs you want to drive through. [01:29] As long as you also make sure they're tested, so they don't sit on our pending dashboard for the better part of a year :) [01:39] RAOF, \o/ [01:40] RAOF, the first one is snapd-glib - bug 1699005 [01:40] bug 1699005 in snapd-glib (Ubuntu Zesty) "Update to 1.13" [Wishlist,In progress] https://launchpad.net/bugs/1699005 [01:40] This one should be fairly uncontroversial, since it's mostly about matching functionality with snapd and it wasn't used in any depth until artful. [01:46] “Doesn't support new APIs” is not normally SRU material :) [01:46] But, obviously, snapd-glib is different. [01:46] RAOF, yeah, is there a more paperwork required to make that more palatable? [01:47] I need to write a Wiki page for an SRU exception, but plan to do that for the next SRU [01:47] robert_ancell: You probably want to do somethinrg based on https://wiki.ubuntu.com/SnapdUpdates [02:08] RAOF, so, I need to write that page for this SRU or we can continue without it? [02:08] We can probably continue without it this time. [02:08] Please get one in future, though :) [02:08] Actually, what are the rdepends on snapd-glib in 16.04? and 17.04? [02:10] RAOF, nothing [02:10] Actually, no, gnome-software is using it but only one method call (to access snapd-login-service) === maclin1 is now known as maclin [02:25] My, my, that's a lot of code. [02:26] robert_ancell: So, what's the test plan for snapd-glib? [02:26] RAOF, check that gnome-software still works [02:26] And run the tests inside the source tree [02:26] We might need to have something a little more detailed :) [02:28] The thing you'd need to test in gnome-software is that (a) it starts up, (b) you can try and install a snap without being logged in to your SSO account and it'll prompt you to log in, and (c) once you're logged in you can install a snap? [02:28] Oh, and I guess that being logged in to the SSO account persists. [02:28] ? [02:28] Just that it prompts you when you don't have credentials and can install with them [02:28] The credentials are managed by g-s, so that shouldn't need testing [02:34] OK. Stick that up on the bug, and I'll wave in into -proposed. [02:38] RAOF, as a comment or the test case? [02:38] As the test case (please edit the bug description) [02:40] RAOF, done [02:41] RAOF, I didn't propose this to Yakkety since it's days from EOL. That wont be an issue will it? [02:41] Nope. [02:42] RAOF, OK, next one is updating gnome-software to 3.20.5 in xenial (bug 1703461) [02:42] bug 1703461 in gnome-software (Ubuntu Xenial) "Update to 3.20.5" [Wishlist,Confirmed] https://launchpad.net/bugs/1703461 [02:43] This one is a bit scary, but ultimately I think puts us in a better position than we are currently [02:44] There's a couple of uploads, but 3.20.5-0ubuntu0.16.04.5 should be the right one (reject the others) [02:44] Well, at least you've pre-filed the paperwork for this one (the GNOME standing exception) ☺ [02:44] I shall check gnome-software after lunch. [02:44] RAOF, cool, thanks === marlinc_ is now known as marlinc [06:12] good morning [06:12] morning didrocks [06:19] duflu: on bug #1704050, how do you try to force the wayland session in gdm? I only have one entry "ubuntu", which should prefers X right now [06:20] bug 1704050 in gdm3 (Ubuntu) "[regression] Can't log in to Wayland Gnome sessions at all" [Critical,New] https://launchpad.net/bugs/1704050 [06:20] there is no "ubuntu on wayland" session like in lightdm [06:20] didrocks, There is Ubuntu on Wayland _and_ GNOME on Wayland :) [06:20] duflu: hum, I don't see that one, despite having the session available [06:20] (on disk) [06:20] Also, I'm working on Wayland :) [06:21] duflu: I would like to have that session to reproduce :p [06:21] * duflu uploads ThinkStation to didrocks [06:21] * didrocks waits :) [06:22] duflu: more seriously, do you think there can be some autodetection making my hw blacklisted in some way? [06:22] like does wayland has a helper to test like we had in nux? [06:22] didrocks, I don't know... The only error I get is from X [06:22] Is there a gdm3 log? [06:22] duflu: unrelated to that error [06:23] duflu: just in general, as you are working on wayland, you may know about that detection mechanism [06:23] duflu: yes, in the journal [06:23] * didrocks doesn't see anything related [06:23] didrocks, No like many people I'm just fighting to make things work with Wayland. I don't really *get it* :) [06:24] duflu: it's an interesting side-effect of some code I changed (appending a session before prepending) [06:24] but grrrr that I can't get a wayland option [06:25] didrocks, my artful install is old. Started with Unity7, then added Ubuntu Gnome, then lots of updates... [06:25] same here [06:25] I wonder if the list is built twice in gdm [06:25] and so, there is an index mismatch [06:26] which might explain your bug [06:26] In fact I have 8 shells to choose from :) [06:26] :p [06:27] Oops, no errors logged because I'm using the workaround [06:28] I bet something is assuming in the code to have wayland sessions before X one on their list [06:29] Oh I see. Yes maybe something is assuming the array order never changes [06:31] mind running with debug enabled? https://help.gnome.org/admin/gdm/stable/troubleshooting.html.en [06:31] OK, will reproduce and check for logs in a sec [06:31] as I can't reproduce as not having that session listed, I would like to keep the same logic of appending wayland sessions and fixing the other part of the code making a wrong assumption [06:33] didrocks, Jul 13 14:31:15 haz gdm3[1046]: Cannot find a command for specified session: ubuntu-wayland [06:34] hum, anything else more helpful in debug mode? [06:35] I can try a patch in the dark and push to a ppa for you to test [06:38] didrocks, lots in debug: https://launchpadlibrarian.net/328568645/gdmfail-debug-log.txt [06:38] Jul 13 14:35:32 haz gdm3[1050]: GdmSession: File 'ubuntu-wayland.desktop' not found: Valid key file could not be found in search dirs [06:38] so, changing the session order hide it [06:39] you have only one desktop file with that name on your disk, correct? [06:39] under /usr/share/wayland-sessions/ubuntu-wayland.desktop [06:39] didrocks, $ ls -l /usr/share/wayland-sessions [06:39] total 12 [06:39] -rw-r--r-- 1 root root 218 Jul 12 18:18 gnome-wayland.desktop [06:39] -rw-r--r-- 1 root root 221 Jul 12 18:18 ubuntu-wayland.desktop [06:39] -rw-r--r-- 1 root root 126 Feb 5 17:57 weston.desktop [06:39] yep, all looks good [06:40] so, libgdm reads in revert orders [06:40] I wonder if that's this list which is unsync [06:40] If it's an array, maybe something is already assuming the order is fixed? So prepending is dangerous [06:41] duflu: it's the contrary [06:41] duflu: it was prepending [06:41] we are appending [06:42] https://launchpad.net/~didrocks/+archive/ubuntu/ppa/+build/13081407 is a stab in the dark, continuing to look meanwhlie [06:45] hum [06:45] with the older version [06:45] when you log into the "ubuntu" session (X one) [06:45] was it really starting X? [06:45] or wayland? [06:45] didrocks, does appending break the 0 element on the end? https://developer.gnome.org/glib/stable/glib-Arrays.html#g-array-new [06:46] duflu: I've used glist appending for years, no, it does the right thing and shift the 0 element on the end [06:47] duflu: but yeah, do you mind trying this? ^ [06:47] so downgrade to working gdm [06:47] try the ubuntu session [06:47] check if it's running wayland [06:47] (I bet it does) [06:47] where you expect X [06:47] didrocks, failed to build? [06:48] duflu: yeah, let's not worry about the small patch right now [06:48] just downgrade to -5 [06:48] * duflu reboots to check it's really X [06:48] I'm pretty sure I understand the issue [06:48] duflu: The gGNOME on Wayland session comes from the gnome-session binary package [06:48] Yes it's really Xorg [06:49] ah, so scratch that [06:49] or maybe… it has smart checks in that way [06:49] I had the same problem with "Ubuntu on Wayland" not working today, but I had a kernel upgrade and more and hadn't tried isolating the cause yet [06:50] jbicha, bisected to gdm already [06:50] jbicha, https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1704050/comments/1 [06:50] Ubuntu bug 1704050 in gdm3 (Ubuntu) "[regression] Can't log in to Wayland Gnome sessions at all" [Critical,Confirmed] [06:50] jbicha: the issue is the append instead of prepend [06:50] search_dirs = get_system_session_dirs (self), [06:50] thanks [06:51] no debug to know what are search_dirs (it should be the list with Xorg, then wayland) [06:51] speaking of gdm, did you see Security's response at LP: #1686393 :( [06:51] Launchpad bug 1686393 in gdm3 (Ubuntu) "[MIR] gdm3" [High,Fix released] https://launchpad.net/bugs/1686393 [06:51] That said I shall reapply the workaround to reconfirm [06:52] duflu: ah, you are not in -5? [06:52] Yep, 3.24.2-1ubuntu5 works fine [06:52] and if you log into ubuntu [06:52] you have the X session [06:52] didrocks, my above debugging was from 7 (where it's broken) [06:52] with that gdm version? ^ [06:52] ahhhh [06:52] please try with 5 [06:53] didrocks, OK I have 5. Retry what part? [06:53] log into the ubuntu session (the X one) [06:53] check it's running X and not wayland [06:53] jbicha: do you know what checks are done on gdm startup? I don't have any wayland session available on my machine [06:54] didrocks, Yes Ubuntu is 'Xorg' [06:54] in 5 [06:54] ok, so definitively not that theory [06:54] thanks duflu [06:54] ok, let's revert the order and set a default session for now [06:54] sounds like the sanest [06:54] Weird. It looks like private information within that function. Order wouldn't matter [06:55] yeah [06:55] that's why I'm really puzzled [06:55] didrocks: do you want to merge this in when you do the upload? https://code.launchpad.net/~willcooke/gdm/ubuntu/+merge/327165 [06:56] jbicha: maybe will should first fix the changelog conflicts [06:56] that would be a good exercise for him :) [06:57] duflu: the issue is from https://github.com/GNOME/gdm/blob/master/daemon/gdm-session.c#L401 [06:57] it can't find that ubuntu-wayland.desktop here [06:57] which is the list we just build… [06:57] with append instead of prepend [06:58] stop changing the changelog then ;) but yeah, his changelog does need cleaning up :) [06:58] Also willcooke has a pending proposal that will conflict [06:59] Oh, you mean that one [06:59] the 2 others parts of the code depending on #ENABLE_USER_DISPLAY_SERVER doesn't seem related to anything order related [06:59] (so meaning: if people build with that disabled, maybe they face the same bug) [06:59] duflu: do you have time for a quick local test + build? [07:00] I would be interested in printing out search_dirs after the line I pointed ^ to know the list content [07:00] Sure, I've been distracted from my actual work for days already :) [07:01] duflu: keep me posted once you get that value [07:03] Ah, I forgot I hadn't answered some upstream questions yet. Even more behind [07:04] In other news: Is there a reason why the Gnome Software icon is different between artful machines? Is that an upgrade issue ? [07:04] (so it's only changed in fresh installs) [07:05] duflu: like fresh install has an ubuntu icon, but not on upgrade, it's the Gnome one? [07:05] Yeah [07:05] duflu: XDG_DATA_DIRS is correct and I have an impacted machine [07:05] duflu: for me, I get the Ubuntu Software branding in 'Ubuntu' but GNOME Software in 'Ubuntu on Wayland' [07:05] so need to look at it, I filed a bug for this already [07:05] jbicha: yeah, that's expected (and we should fix that) [07:05] OK, no problem. [07:06] In other other news: Whose idea was it to design buttons that don't look like buttons in Gnome? [07:06] I wonder if G_N_ELEMENTS() isn't puzzled by NULL though [07:07] didrocks, It's a null terminated array so maybe that flag tells it to count the size manually === [PARTY]Varka3rd is now known as [PARTY]Varka [07:07] yeah [07:07] Interesting [07:07] mind priniting its result? [07:08] (the G_N_ELEMENTS() count) [07:08] I really wonder if that code branch never worked [07:08] G_N_ELEMENTS() is compile time: it doesn't know what is stored in the array [07:09] Yeah that's for C arrays not GArray [07:09] #define G_N_ELEMENTS(arr) (sizeof (arr) / sizeof ((arr)[0])) [07:09] good morning desktoppers [07:09] hey oSoMoN, jamesh [07:09] Hi oSoMoN [07:09] salut didrocks, hey duflu, jamesh [07:14] didrocks, actually I can tell you just by reading it that use if G_N_ELEMENTS is wrong [07:14] It won't find the wayland entry because that's after a NULL [07:15] Problem solved [07:15] duflu: wait, the count doesn't have NULL, so only the number of elements [07:15] as it's based on sizeof(arr)/sizeof(elem) [07:15] didrocks, yeah NULL gets appended, before wayland [07:16] So only prepending works [07:16] hum [07:16] G_N_ELEMENTS (x_search_dirs) is 4 [07:16] no? [07:16] Should be 5 (I am not up to debugging yet) [07:16] as NULL takes 0? [07:16] (in term of size) [07:16] didrocks, No it's the number of pointers in the array, hence 5 [07:16] ohhhh [07:17] yeah, it's pointers ofc [07:17] so yeah, NULL is set in search_array [07:17] Good news - didrocks did the right thing and was fooled by bad code [07:17] and then, wayland is appended [07:17] Maybe [07:17] yeah, that code branch never worked [07:18] tell me once you get that debug printed out :p [07:18] Actually the code is good, but it was nonobvious that append would not work [07:18] well [07:18] look at the original code [07:18] https://github.com/GNOME/gdm/blob/master/daemon/gdm-session.c#L360 [07:18] if you define ENABLE_WAYLAND_SUPPORT and not define ENABLE_USER_DISPLAY_SERVER, you are in that case [07:18] (which doesn't work) [07:19] didrocks, Oh yes, the bug already existed [07:19] So indeed "Good news - didrocks did the right thing and was fooled by bad code" [07:19] :p [07:20] didrocks, I think someone can fix that without further testing...? [07:20] yeah [07:21] do you want to do it? I can do it, just don't want to steal your commit :) [07:21] didrocks, well I can test and verify it for sure... no problem [07:21] and praise when reported upstream [07:21] ;) [07:37] duflu: tell me once you get your patch tested and I'll just sponsor it [07:38] didrocks, yeah I'll do it today. Probably Ubuntu before upstream tho [07:38] duflu: sounds good [07:38] duflu: just open a bug in bugzilla and reference it on the quilt patch [07:39] (even if I think next time we touch that patch we'll just inverse the logic back to favor wayland sessions) [08:02] morning [08:03] Morning Laney didrocks duflu oSoMoN jamesh [08:03] hey Laney, flexiondotorg [08:03] good morning Laney, flexiondotorg [08:04] Morning flexiondotorg [08:04] Morning Laney [08:06] hey flexiondotorg hey didrocks hey oSoMoN hey duflu [08:06] how's it going? [08:06] not too bad round here [08:06] good, started with some interesting gdm bug :p [08:07] oh yeah [08:07] what's that? [08:07] but everything is under control once someone has a session to test :) [08:07] Laney: basically, the branch with "append" instead of "prepend" never worked [08:07] the bottom of it is that the previous list we append to already has a NULL pointer [08:07] so, the wayland session was never found [08:08] (patch to get upstream, even if I don't think anyone is building with the 2 #ifdef condition) [08:08] ah right [08:08] morning all [08:11] hey willcooke [08:11] * didrocks needs to add a new expection in the seed for a -dbg package [08:13] thunderbird pulls thunderbird-gnome-support-dbg in main, which brings thunderbird-gnome-support, which brings indicator-messages, which pulls libindicator-messaging-menu, which … [08:17] hey Laney willcooke [08:17] or and other desktopers, looks like I was too busy to chat on other channels and forgot to say hi here [08:18] Morning willcooke seb128 [08:19] hey flexiondotorg [08:21] salut seb128 [08:21] good morning willcooke === JanC_ is now known as JanC [08:21] How's it going oSoMoN? I'm reading good comments about the new LO snap - great stuff! [08:22] willcooke, doing good, thanks! trying to wrap up everything that’s on my list before I leave for holidays [08:22] hey desktopers [08:22] hey ricotz [08:23] hey oSoMoN [08:24] duflu, still around? https://bugs.launchpad.net/ubuntu/+source/sound-theme-freedesktop/+bug/1703946 [08:24] Ubuntu bug 1703946 in sound-theme-freedesktop (Ubuntu) "Dedicated mono test sound is unavailable" [Undecided,New] [08:24] Do I just update the changelog to say artful? is it that simple? [08:26] willcooke, yes I think so [08:27] it's quite normal for the uploader to do that too [08:27] ah, good to know. Well, I'm happy to just edit the debdiff and reattach it, if that helps [08:27] I'll attach one with the change anywy, wont take a sec [08:34] Laney: I was wondering that, thanks [08:34] Not like it's checksummed [08:43] didrocks: OK, done. Works too. [08:44] lut oSoMoN [08:45] hey seb128 [08:45] you good? [08:46] duflu: excellent, MP against the ubuntu branch? [08:46] didrocks, https://code.launchpad.net/~vanvugt/gdm/fix-1704050/+merge/327360 [08:47] duflu: does it work with the prepend case still? [08:47] (when you submit it upstream) [08:47] duflu: just: mind opening a bug upstream and reference it from your patch? [08:48] didrocks, already have an upstream bug with the patch [08:48] duflu: just add it on top of debian/patches/Avoid-double-NULL-terminated-array-LP-1704050.patch [08:48] that way, we can track [08:49] didrocks, I was reluctant to reformat as I could not remember the rules. Also upstream already have that patch. The upstream bug is linked in Launchpad now [08:49] duflu: linking in LP is good enough, thanks! [08:49] merging and sponsoring [08:51] hum, no tag and you set it to artful [08:51] * didrocks adds a tag [08:52] didrocks, we were just talking about that. Do you prefer UNRELEASED if not tagged? [08:52] duflu: I generally prefer UNRELEASED not tagged, but not a biggie [08:53] OK then willcooke got it more right [08:53] For reviewer=didrocks [08:53] (so, basically, a separate commit to set it to artful + tag for release) [08:54] didrocks, cool will do in future [08:54] I don't think I usually/ever propose to ubuntu branches [08:54] duflu: no worry! sponsored :) [08:55] So sorry in advance to willcooke. I am making new conflicts in gdm [08:55] ;) [08:56] highly demanded package [08:56] didrocks, I _assume_ and it seems to be true that append does go before the implicit NULL :) [08:57] duflu:yeah, I hope that the prepend case does have an implicit NULL so that your fix is accepted :) [08:58] didrocks, I think prepend will always work regardless of the contents [08:58] I meant, if there is no match and the list isn't ended by NULL… [08:59] didrocks, I think new = {NULL} [09:02] Laney, I'm good thanks, you? [09:03] good! [09:03] went to folk club last night, was funny [09:05] nice [09:06] didrocks, do you recomment UNRELEASED for debdiffs too? [09:06] recommend [09:09] duflu: no, debdiff can be the final thing, ready to sponsor [09:10] Makes sense [09:10] both ways [09:10] the only thing with UNRELEASED in a branch is that the author can ammend, commit separated a fix while still being in the "UNRELEASED" part [09:10] amend* [09:13] Could one of the desktopers merge this please - https://code.launchpad.net/~ubuntu-mate-dev/indicator-session/mate-integration/+merge/325600 [09:13] I've addressed all the feedback and been running a PPA build for several weeks. All good from my point of view. [09:19] didrocks, next time please copy the commit message field :) [09:19] duflu: I didn't commit anything [09:20] duflu: as you set it to "artful" [09:20] OK I'm confused [09:21] The Ubuntu branch history shows my garbage commit messages, but failed to use the formal commit message from the MP [09:21] just bzr pull && bzr push (after adding the tag): the git way when you are on align with master [09:22] aligned* [09:22] didrocks, yeah the bzr way is different. If landing by hand you need to copy the commit message from the MP because it's not in bzr [09:23] duflu: depends on the branch, some prefer to always merge, some not. [09:23] didrocks, fair enough. I'll keep it in mind [09:23] we should just standardize on one way to do it IMHO [09:24] so that you don't have wrong expectations, I see what you wanted… [09:24] (also one tool, not some branches in git vs other in bzr) [09:24] I guess a discussion at the start of a cycle, one all the transitions are over and polish done :) [09:37] L_aney is going to work on the git-for-packaging for Debian pkg-gnome during debconf [09:37] so we might want to look at building on top of that next cycle then [09:37] nice! [09:37] a discussion to have for when we get our team week [09:37] yep [09:57] * duflu -> vegetable peeling fun [10:27] quick reboot [10:29] popey: if you are too impatient to wait for gdm to published in the release pocket, it's in -proposed right now [10:30] a gdm3 upload a day [10:30] keeps the something something away [10:30] ;) [10:30] "bugs" for one someting [10:30] but need another adjective [10:31] am I the only one to get this with wayland + our theme on gnome-terminal: http://imgur.com/a/k48Se ? [10:31] ah, that's the transparency feature [10:31] didrocks, it's a known bug [10:32] it's reported [10:32] ah nice, do you know if our transparency patch is the same than in fedora? [10:32] * didrocks will look [10:32] bug 1650395 [10:32] bug 1650395 in gnome-terminal (Ubuntu) "Large grey border around gnome-terminal shadows when transparency is enabled (wayland)" [Medium,Triaged] https://launchpad.net/bugs/1650395 [10:33] more or less [10:33] interesting that the color is theme-dependent [10:33] * didrocks likes [10:33] but maybe rishi fixed it or something [10:33] my borders are bigger :) [10:33] and we need to pull i again [10:33] yeah, would worth a look [10:33] he's got a repo with it in [10:34] hum, do you want to test it? (I'm not going to do it today, just as you seem to know more where things are…) [10:34] I'm happy next week to have a look otherwise [10:34] https://github.com/debarshiray/gnome-terminal [10:34] * didrocks got a new tab [10:34] thx [10:34] we took that and then added the theme colours stuff [10:35] will look eventually, but not right now [10:35] so whoever gets there first can :-) [10:35] deal! [10:35] :-) [10:36] actually I'll ask him quickly, that'll be easier [11:09] has anyone tried artful on a wired ethternet connection? I can't get it to accept a DHCP address. Its being offered one, but seems to be ignoring it. cyphermox perhaps? [11:14] didrocks: will test! [11:17] popey, be a love and test wired network as well would you? [11:18] I'm on wired right now [11:18] these bits are wires alllllllll the way baby [11:19] mannn [11:19] that means its something on my network. Damn it. [11:19] * willcooke tries a live session [11:31] willcooke: i am on wired [11:32] and it just works? [11:32] curses [11:32] wfm [11:33] * willcooke breaks out tcpdump [11:33] what does "nmcli d" say? [11:33] I know it's requesting a dhcp address, I can see it on the server [11:33] and in syslog [11:34] i haven't setup wifi on this laptop yet [11:34] only wired [11:36] ok, connected to wifi, also works, and doesn't seem to have a problem switching between wired and wifi [11:41] anyone aware of touchpad issues with xenial in the last few weeks? [11:41] mostly with double finger scrolling and syntaptic device [11:43] i haven't. been using a touchpad for the first time in a while these last couple of weeks and it's just as crazy as ever (ie. not a nipple) :D [11:48] andyrock, what sort of issues? [12:01] popey: it's impossible to stay away from nipples :-D [12:01] I can't too [12:01] * Laney titters [12:01] Laney: you should get a thinkpad or at least a thinkpad BT keyboard to love it more :-D [12:04] mpt: hey, remember that change for u-c-c, we defined the graphics levels as "High" and "Low", but I think we should also add the "auto-detect" mode... As by default we use that way, while we use the setting only when set [12:04] are you ok with that? [12:55] FINALLY indicator-messages off main [13:05] didrocks, well done [13:08] seb128: upstreams says "no, fix your tools" :) https://bugzilla.gnome.org/show_bug.cgi?id=784877 [13:08] Gnome bug 784877 in general "Don't use variable for gettext package name" [Minor,Resolved: wontfix] [13:17] seb128: seems like it does not work after login (the mouse) [13:17] or a new created user session it works fine [13:17] didrocks: nice work fixing gdm :) [13:17] probably not a but (just a misconfiguration) [13:17]