=== simonquigley is now known as tsimonq2 [06:15] Good morning [06:30] Morning jibel [06:31] Hi duflu [06:42] good morning === Mirv_ is now known as Mirv === pstolowski|afk is now known as pstolowski [07:03] good morning desktopers [07:06] Hi didrocks, seb128 [07:07] hey duflu, how are you? [07:07] seb128, dazed and confused. That massive memory leak I was working on is gone. So now switching tasks [07:07] You? [07:10] hey duflu, re seb128 [07:11] * duflu wanders off for a little [07:12] duflu, some update resolved it? [07:24] seb128, no I can't see any updates [07:24] weird :/ [07:25] I am seeing "smaller" leaks of 0.5-1MB but those are much smaller than my original focus [07:27] And then sncf emails me to try and sell Summer holidays in France. Sounds good [07:27] morning [07:27] Morning Nafallo [07:28] would it be a bug if gnome-initial-setup show the installed app in the suggested apps as well? [07:30] Nafallo, I don't know, that's a question for robert_ancell [07:31] user experience? *shrugs* [07:31] Nafallo, that was raised but I'm not sure what the conclusion was. [07:31] I just found it odd :-) [07:31] I'll let design decide - note of course you tend to only see this after an install, so you're not likely to see any overlaps. [07:32] unless I pre-seed stuff ;-) [07:32] indeed :) [07:33] other than that it looks good. couldn't test the livepatch stuff, since I've already used up my member entitlements :-) [07:34] hehe. metrics.ubuntu.com doesn't exist yet :-) [07:37] yep [07:40] good morning desktoppers [07:41] salut oSoMoN [07:41] lut oSoMoN, en forme ? [07:41] salut didrocks, seb128 [07:42] daughter was sick so that was a tough night, but it's over :) [07:42] how are you guys? [07:43] I'm a bit sick but otherwise good [07:43] getting in better shape little by little :) [07:53] Hi oSoMoN [07:53] hey duflu [08:02] hey [08:03] hey Laney [08:06] good morning Laney [08:07] good morning u.k [08:07] hey willcooke Laney [08:07] hey oSoMoN didrocks seb128 [08:07] what's new? [08:08] it's cold & windy, not supposed to be like this in april! [08:08] sun here! [08:08] weekend should be alright here. between 2-10°. [08:08] camping weather :-) [08:10] morning. [08:10] hey hey willcooke [08:10] blurgh [08:11] hey willcooke [08:12] hey willcooke :-) === cpaelzer_ is now known as cpaelzer [08:14] Morning willcooke [08:14] and morning Laney === upmine is now known as niko [10:18] willcooke, I'm all done with the review of beta 2. 2 majors issues: 1. Cannot upgrade from ubiquity 2. No screen reader support [10:18] major* [10:18] jibel: \o/ I guess 2. isn't a blocker for final beta as well? [10:21] sil2100, none of them are [10:23] Yeah, I knew 1. wasn't, just making sure about 2. [10:27] thanks a lot jibel, good work [10:28] jibel, screenread from the installer? [10:29] willcooke, yes, the shortcut to activate it doesn't work and there is no indicator either in the panel [10:29] bug 1741690 [10:29] bug 1741690 in ubiquity (Ubuntu) "You can't enable the Orca screen reader until after you click "Try Ubuntu" on Ubuntu bionic" [Critical,Confirmed] https://launchpad.net/bugs/1741690 [10:31] Laney, re fontconfig, would it work to touch the font files when building the filesystem to remove the nanoseconds then re-run fc-cache? [10:31] instead of modifying fontconfig [10:32] and it'd work on all the flavors [10:47] jibel: I'm pretty sure my fix fixes that case too, as it ignores nsec == 0 - so if the argument is that things end up with nsec = 0 on the target then a regen won't be triggered. [10:47] Not sure why an LP specific fix is better? [10:48] Also I don't know offhand how to do that ... [10:49] * Laney downloads a xubuntu iso [10:50] not that I can get on cdimage for some reason [10:55] just having some drama with a leaning fence [11:12] jibel, would be interesting to try the screenreader with the fontconfig fix [11:15] seb128: it was happening in 17.10 from what we got told, so few chances… [11:15] weird, I though we fixed it in N-Y [11:19] maybe the bug was simply not closed? [11:19] could be, but it's buggy for sure now in bionic [11:19] but IIRC I asked the reporter to try a 17.10 iso and he confirmed it didn't work [11:20] I'm going to wait for that fontconfig fix in any case [11:20] yeah [11:23] seb128, yes, I suppose the shortcut don't work because gsd services are timeout, like the terminal shortcut for example [11:24] right [11:37] sil2100, there is something else, there is no maybe-ubiquity mode [11:38] could someone with a bios machine boot latest bionic iso and tell if they see the try/install screen? [11:39] I just have an uefi machine and in a VM it always shows me the syslinux menu [11:39] I do under virtualbox [11:39] I can try in a bit [11:40] seb128, the boot menu is not displayed in vbox and it boots to ubiquity-dm? [11:40] I get the menu but that iso might be some days old [11:40] is that a recent issue? [11:43] idk, I rarely boot without pressing a key [11:45] jibel, dunno if the comment from Steve on https://code.launchpad.net/~vorlon/ubiquity/getties-and-ubiquity-dm/+merge/342716 has to do with your issue [11:45] or if he was trying to debug that [11:46] sounds the second to me [11:46] right, I was not suggesting the code change there has to do the issue [11:47] rather I wonder if his " OTOH it doesn't look like the current Ubuntu ISO does maybe-ubiquity on its own" is stating the same problem [11:54] probably an issue with libvirt. With an older version it boots to ubiquity-dm and does not display syslinux === pstolowski is now known as pstolowski|lunch [12:05] jibel: what would you want to be different from libvirt in this case? [12:11] cpaelzer, it's like the guest is receiving a keystroke on boot. [12:11] which it should not [12:12] cpaelzer, I'm upgrading the second machine to the same version of libvirt to check if it's the problem [12:17] odd [12:18] jibel: libvirt won't send keystrokes, maybe some defaults on the console changed which lead to this new behavior [12:18] jibel: once you spot what really happens let me know [12:18] libvirt does not interact with the guest console [12:18] it only does with the qemu monitor [12:19] jibel: is that on graphical boot (e.g. through virt-manager) or a pure console guest (e.g. virsh console)? [12:23] cpaelzer, it's a graphical boot. I'll dig further and let you know if I find anything. [12:23] pfff - who wants graphic :-) [12:23] jibel: looking forward to hear from you [12:30] jibel Home directory encryption is absent from 18.04. Intentional? [12:30] flexiondotorg, yes, it's been removed. [12:31] flexiondotorg, ubiquity 18.04.3 [12:31] * Remove the encrypted userdir option from the gtk installer since [12:31] ecryptfs is going to move to universe. The keep the base code in case [12:31] the option is added back later using another technology. [12:31] Hmmm, how are OEM installs going to offer data security without home encryption? [12:31] willcooke: ^ [12:36] people will need to manual setup whole disk encryption [12:36] manually [12:37] or move their home dir to another partition and enable ext4 encrpytion [12:37] flexiondotorg, disk encryption [12:38] Fine for individuals. But what about OEMs shipping computers with OME setup? [12:38] SED drives? just throwing it out there. [12:38] flexiondotorg, talk to the security team, we are not the ones who decided on that change [12:39] flexiondotorg, they can enable encryption after OEM setup is complete in the same way as anyone [12:39] I don't follow. [12:39] FDE is a very early install step, not typically something done at the end. [12:40] there's been some work trying to get cryptsetup to encrypt unencrypted installation. I haven't dared tried it yet. [12:41] flexiondotorg: possibly encrypt it with a passphrase file on /boot or somewhere and then write software that asks the user for a password or two before removing the keyfile? [12:42] anyway. sort of offtopic for this channel. [12:43] If a vendor perform an OEM install to, ship on laptops sold to customers, and choose FDE the encryption key is known by them. [12:44] hence why you remove the keyfile after setting the user password(s) [12:48] flexiondotorg, I'm not really up to speed with what the encryption options are. Best to ask the security team [12:48] OK === pstolowski|lunch is now known as pstolowski [13:10] cpaelzer, bug 1761497, not urgent [13:10] bug 1761497 in virt-manager (Ubuntu) "syslinux menu is displayed when a desktop ISO is powered from virt-manager" [Undecided,New] https://launchpad.net/bugs/1761497 [13:10] it doesn't happen from the command line [13:11] jibel: TBH I'm not sure what to do on this :-/ [13:11] do I read this correctly "not all the physical machines running the same version of libvirt and qemu exhibit this behaviour" [13:12] that even for you only "some" systems show this effect [13:12] even being on the same versions? [13:13] cpaelzer, yes, it's weird. I uninstalled/reinstalled the vm stack and no change. The machines I tried have different graphics chipset [13:13] jibel: well the virtual graphic should not change [13:14] jibel: when you say "started with virsh start" does that mean with a pure text console? [13:14] like with "virsh start --console" [13:14] it is not blocking anything, just annoying [13:15] cpaelzer, yes, virsh start [13:16] jibel: but from commandline still gives you a graphical view? [13:17] ubiquity-dm was the nice graphical prompt right? [13:18] cpaelzer, ubiquity-dm is the ubiquity dialog with a list of languages and the options try or install ubuntu. It's the first thing a user should see if he doesn't press key during boot [13:19] cpaelzer, otherwise if you press a key on the aubergine screen with a small keyboard icon at the bottom it proposes several other options and it's in text mode. [13:20] cpaelzer, that's what I get even without pressing any key [13:20] but really don't spend time on this, I can use the command line [13:33] willcooke, I'm going to mark ubuntu desktop ready to release. Any objection? [13:34] jibel, +1 thanks [13:35] so ecryptfs-utils has dropped off all the isos, meaning the encrypt home folder checkbox for KDE front end can't work? [13:35] acheronuk, right, if you want to keep providing that feature I guess you need to make -kde depends on it [13:36] seb128: the bug says it's buggy. problematically so? if you know that is [13:37] I don't want to add something back that is likely not supportable [13:37] acheronuk, well, we removed the depends so up to you to remove the option as well or add back the depends [13:37] I'll investigate. thanks [13:39] yw [13:47] seb128: Mystery with deactivated templates resolved. Gabor hasn't replied yet, though. [13:47] GunnarHj,hey, I saw that! [13:49] seb128: Do you have an idea off hand which packages in universe make use of the language packs? [13:49] no [13:49] :( [13:49] why? [13:49] to know if there are enabled? [13:50] seb128: If their templates were deactivated, the translations will disappear for the users at next full langpack update. [13:53] GunnarHj, then we can diff the langpacks, enable those back and do another langpack update [13:53] or ask on #ubuntu-devel if somebody knows how to query for packages using that tag in their control [13:54] seb128: The latter sounds promising. [13:56] GunnarHj: too bad we don't have something like for Ubuntu: https://codesearch.debian.net/search?q=X-Ubuntu-Use-Langpack [13:57] L_aney set up one a while ago but I guess that didn't get maintained [13:57] jbicha: Wow! One first thing to do is to check whether the templates for those packages are up and running. [13:58] GunnarHj: that list is very incomplete for what you want though [13:58] it won't have any of the Unity packages! [13:58] jdstrand used to have a scrip that does grep through the archive, unsure if he still has it [13:59] jbicha: I know. But it's a starter. And we are dealing with Unity separately anyway. [14:46] jbicha, seb128: Could bring back template for evolution by help of that Debian search. [14:46] GunnarHj: ? [14:47] I thought evolution was already in the langpacks [14:47] that was our intent at least! [14:47] jbicha: It is. The problem is (was) that the template got deactivated by mistake. [14:48] I have been confused by the langpacks for years. Guess I'm still going to be confused :( [14:49] jbicha: For context on the present archive wide issue: [14:49] https://lists.ubuntu.com/archives/ubuntu-translators/2018-April/007474.html [14:49] I seem to remember there being a text file or something [14:51] How do we dig deeper into Laney's memory? :) [15:07] seb128: Have they replaced ++U with ++E because that emoji thing, even if it's not compliant with the standard? [15:07] GunnarHj, the details are in the bugs I pointed out [15:07] but yeah basically [15:08] seb128: Sounds weird. I'll switch to bionic and test it to start with. [15:09] GunnarHj, well it's not going to tell you much, don't bother, I'm handling it [15:09] seb128: Are you going to bring Ctrl+Shift+U back then? [15:28] didrocks, go doesn't make small C libraries! libsysmetrics is almost as big as libgtk (5.7M vs 7.1M) [15:28] seb128: yeah, there is the Go runtime in it [15:29] (and the stdlib) [15:30] k, makes sense [15:31] we can even embed the python runtime inside the Go runtime if you want more inception :p [15:31] (but then python is single-threaded due to their gold locker) [15:54] tempted to upload my fontconfig proposed patch [15:54] and replace / refine it with upstream comments if we get any [15:55] any opinions? [15:56] +1 [15:56] I think it's the pragmatic thing to do [15:56] and we need that issue resolved to see if other problems go away like keybinding, screen reader, etc [15:56] also having a not-so-slow iso boot might make other work easier [15:57] the fix also seems to make sense, it's a bit hackish but it should do the job and I don't see a potential problem due to it [15:57] ok then [15:57] thx [16:09] yeah, sounds the best approach [16:26] done DONE [16:26] stupid sun revealed how dirty my screen is [17:14] night all [17:25] hey Laney can you publish https://bileto.ubuntu.com/#/ticket/3227 please? [17:25] Trevinho: I'm just off, can look tomorrow if nobody beats me [17:25] see you! [17:26] Laney: ah, ok no problem [17:26] good night [17:26] kenvandine[m][m]: maybe? ^ [18:24] seb128: can we use https://salsa.debian.org/gnome-team/gnome-initial-setup for packaging since there never was a bzr branch for it? [18:24] oh, I mean I guess we should use the LP thing we did for gvfs [18:25] https://code.launchpad.net/~ubuntu-desktop/ubuntu/+source/gvfs/+git/gvfs [18:26] there is an outdated ubuntu/bionic branch in Debian I guess I can remove (outdated because I figured out how to sync our packaging earlier) [18:26] and that branch was there because it was allowed to use salsa for universe packages that didn't have LP packaging branches [18:59] jbicha, if the people who work on it know the workflow and agree, I'm not going to contribute fixes via the vcs this cycle if you do that though [19:00] were you planning to submit fixes for gnome-initial-setup? ;) [19:00] we can sponsor fixes this month if you do :) [19:00] well I did a few in the ppa already [19:03] jbicha, I'm just going to dput changes I need don't worry, then somebody can sort out the vcs [19:05] jbicha, check with robert_ancell that he's fine with that workflow though since he's the one doing the active work atm [19:05] yup, I was going to chat with him [19:05] thx [19:05] but time zones :) [19:06] right, well he should be there in a few [19:06] but you won't be hopefully! [19:06] :) [19:06] don't worry I don't plan to join that discussion [19:07] I already stated my position of disrupting workflow at this point of the cycle === pstolowski is now known as pstolowski|afk [21:07] robert_ancell: good morning. I tried out the ubuntu-welcome ppa :) [21:07] jbicha, thanks! [21:08] when are we planning to upload it to bionic? [21:08] jbicha, today [21:09] did you want to use a Vcs for the packaging? [21:09] jbicha, any proposals? [21:11] you saw Lan_ey's gvfs packaging? a ubuntu/master git branch in the LP ubuntu-desktop team derived from Salsa [21:11] seb128 has said he won't submit any changes to git branches this month but we can adapt his debdiffs or whatever [21:13] jbicha, I haven't seen that but seems reasonable [21:14] you weren't at his talk in Budapest? [21:15] jbicha: we didn't decide much how to proceed with that [21:15] I'm totally in with moving all our bzr branches there [21:16] as it really improves my workflow [21:16] so I'll be happy to help with the transition [21:16] Trevinho: it breaks seb's workflow though! :| [21:17] jbicha: well, we can always make a git hook that will transfer it over bzr xD [21:17] But really, using git is really easy and it won't take more than 1 hour to get used to this system [21:17] we should also decide what platform to use [21:18] if he has never used git before, it might take some time as git has odd workflow issues [21:18] I think most of us have been using git for years by now though [21:18] yeah... [21:19] i like bzr for some ways, but for sure not the way I can handle the deb branches we have [21:19] it makes incredibly hard to just add a patch... [21:19] the Debian GNOME workflow is a little more complicated than basic git-buildpackage (and using LP makes it just a bit more complicated) but it's ok [21:19] That's why I already was use git-buildpackage + bzr for my purposes [21:20] not having access to the repo makes things harder to manage [21:20] while having a repo I can write to, or PR against with all the code too, would make things way easier [21:21] Trevinho: have you done git merge proposals using LP before? [21:21] jbicha: a few times [21:21] is the UI ok compared to gitlab? [21:22] jbicha: well, gitlab is better to be fair [21:22] jbicha: I'd use the GNOME one if possible, but not sure they like that :) [21:22] launchpad is ok anyway, it has better points for prerequisites for example [21:23] but the ui isn't nice as Gitlab has now [21:23] it's funny because there are 3 of us that have opinions on where we want the Ubuntu packaging and they're all different [21:23] jbicha: the way I was doing :-D https://gitlab.gnome.org/3v1n0/nautilus/commits/ubuntu-3-26+patches [21:23] bzr + git [21:23] you want it to be on GNOME, I want it to be on Debian, and Lan_ey wants it to be on LP [21:23] ahaha [21:23] well, on Debian, if I can't do much with salsa as a guest, I'd prefer no :-) [21:24] It's true that launchpad is the natural places [21:24] I mean we can always have multi-repos around but we should select one for PRs [21:24] we can add users to our Debian GNOME repos [21:24] and in general where's gitlab is better [21:24] and with gitlab you can always fork and propose merges (like you can on LP or github) [21:25] anyway, the host and review system is just secondary [21:25] ok, ok... that's fair for me [21:25] Having it in debian, I don't know... We can, if they prefer like that to eventually cherry-pick from us [21:25] not sure [21:25] I've not hard requireements [21:26] other than that I don't want bzr for this again .) [21:26] like 80+% of Debian GNOME stuff is in sync between Debian & Ubuntu so we already need to use Salsa to properly contribute during much of our devel cycle [21:27] (the percentage is rough depending on what exactly you're counting) [21:28] What I need as doing this mix of upstream and downstream work is.. Something I can quickly edit code, or cherry pick from soemthing I already did, and generate debian patches [21:28] and that's what gbd does [21:28] A part that I can't pronunciate that f***ing tool :-D [21:29] gbp pq* [21:29] I know what it means, but it's the most annoying thing to remember and say ever! [21:32] jabberwocky [21:33] the Debian GNOME workflow sort of forces you to have local git remotes [21:34] so something like gbp pq import; git fetch gnome (if necessary); git cherry-pick foo1234; gbp pq export is pretty nice [21:40] robert_ancell: if you weren't at Lan_ey's talk at Budapest, maybe it's easier to just not use a vcs for gnome-initial-setup this month [21:41] jbicha, I was but I'll probably just upload directly this time and then we can migrate it in later [21:42] robert_ancell: I can set up the repo if needed, but I'll let you decide what we do since I expect you're doing most of the packaging this month :) [21:43] jbicha, if you set it up I'm happy to push to it. You'll probably have to give me a cheat sheet to remind me of the commands :) [21:43] cheat sheet is at https://wiki.debian.org/Gnome/Git [21:43] this is an important divergence from Debian: https://git.launchpad.net/~ubuntu-desktop/ubuntu/+source/gvfs/commit/?id=2eb784c29 [21:44] you have to use gbp buildpackage --git-tag-only to get the nice ubuntu/3.28.1ubuntu1 tag since 'debcommit -r' won't give that to you [21:45] and it needs a separate remote for Launchpad so that gets a bit more complicated with pulling and pushing [21:45] it kind of sounds complicated to me even! so I'm ok with you not wanting to mess with it right now [21:48] robert_ancell: by the way, Initial Setup uses the gnome-control-center icon basically. Would you be interested in a patch for the .desktop to use Icon=ubuntu-logo-icon instead? [21:48] jbicha, I don't have a strong opinion on it - I guess it makes sense but should we be able to keep the existing behaviour if not running in standard Ubuntu? [21:50] we don't currently keep the existing pages for non-standard Ubuntu, right? [21:51] we could ship a /usr/share/ubuntu/applications/gnome-initial-setup.desktop to override in the Ubuntu session (like we do with gnome-software and imagemagick) [21:51] jbicha, we do - it changes the page list based on XDG_CURRENT_DESKTOP [21:51] we could ship that override in ubuntu-settings to keep the gnome-initial-setup packaging cleaner with respect to Debian and upstream [21:52] oh I hadn't looked at the repo recently :) [21:52] or closely [21:53] np, it's all a bit of a last minute hack [21:53] let me work on a .desktop to put in ubuntu-settings then [21:53] BOOM, uploaded as 3.28.0-2ubuntu1 [21:53] Expected to be fixed a few times before release [21:54] robert_ancell: by the way, I filed 3 bugs today: https://bugs.launchpad.net/ubuntu/+source/gnome-initial-setup [21:55] jbicha, is 1614816 the reason that the language selector is disabled? [21:55] yes [21:55] is that a problem? [21:56] complicated by LP: #1631750 [21:56] Launchpad bug 1631750 in gnome-control-center (Ubuntu) "Language installation doesn't work in Ubuntu GNOME 16.10 Settings app" [High,Triaged] https://launchpad.net/bugs/1631750 [21:56] jbicha, no, I was just wondering why it was disabled. [21:58] it's disabled in Fedora 28 too actually [22:00] there's two additional issues: it duplicates a question asked in the installer and if gnome-initial-setup is run in its existing-user mode, it can't actually change the system language until log out and log back in [22:00] existing-user is used unless we don't have ubiquity create the first user [22:01] ah [22:02] hmm, I'm guessing we can't actually do the .desktop override in ubuntu-settings easily because we don't have ubuntu-settiings set up to be translatable [22:03] jbicha, thanks for the bugs! [22:03] I wonder how hard it would be… [22:07] I wish I could edit LP bug comments... [22:10] to be more specific, Fedora disables the language selector in existing user mode only because of those 2 issues [22:12] see vendor.conf which depends on reduce-initial-setup-redundancy.patch at https://src.fedoraproject.org/rpms/gnome-initial-setup/tree/master [22:12] that way Initial Setup can be used as Fedora's OEM mode too [22:14] Pop!_OS is the only major distro I'm aware of that really uses gnome-initial-setup's new user mode [23:42] robert_ancell: I pushed the .desktop with new name and icon to https://code.launchpad.net/~ubuntu-desktop/+junk/ubuntu-settings [23:43] we don't have to wait for the gnome-initial-setup MIR to upload since that .desktop has NoDisplay=true anyway [23:43] jbicha, nice [23:43] give it a try and let me know what you think :) [23:45] mean while I verified that gnome-initial-setup keeps the old pages on GNOME. Interesting. [23:52] jbicha, open for feedback on that - it wasn't clear to me exactly what we should do / I don't want to take away from those who want the vanilla experience. [23:52] If it was just a few small changes I think it would be OK, but since we replace essentially all the pages... [23:52] I'll check with darkxst next week [23:52] ta [23:56] I guess we could ask Budgie and Unity guys too since they use GOA. It's easy to whitelist additional desktops