[06:35] good morning [06:38] * tsimonq2 waves to didrocks [06:39] hey tsimonq2 [06:40] hey desktopers [06:40] hey ricotz [06:40] didrocks, hi [06:43] Morning didrocks, tsimonq2, ricotz [06:43] hey duflu [06:48] o/ ricotz duflu [08:03] yo [08:04] hey Laney, how are you? [08:04] hey willcooke [08:04] good morning desktopers [08:06] morning [08:07] huh, I've never seen this before: https://github.com/CanonicalLtd/desktop-design/blob/master/Progress/spinner.svg [08:10] willcooke, where is it supposed to be used? [08:11] Morning Laney, seb128, willcooke [08:12] seb128, I don't really know. It looks like m_pt is uploading a lot of design assets and docs from all over the place in to github, and that was one file I clicked on by random [08:12] hey duflu, how has your day been so far? [08:12] willcooke, k, never saw it either [08:13] seb128, painful, but only because I'm trying to fix totem again [08:13] did you make any progress figuring out what regressed? [08:13] duflu, I tried with 18.04 yesterday and I /think/ it is a bit juddery. I wanted to install 17.10 and try that but I needed that machine yesterday [08:13] I can reinstall it today I think [08:14] seb128, seems to be GLX-specific, so Xorg sessions mostly. And I am looking at a realtime profile showing it spending 60% of its time waiting in synchronization calls [08:14] It's also confused by 18.04 using upstream's gst-vaapi code (17.10 used my own patch). But I hope that's not an issue [08:15] willcooke, you mean video is juddery? [08:15] duflu, yeah [08:15] hey seb128, got a sore throat but otherwise ok [08:15] you? [08:15] moin duflu willcooke [08:15] word Laney [08:16] BTW, that xwayland crash - do I need to do anything else, or is it understood now? [08:16] good morning desktoppers [08:17] willcooke, Probably nothing. But maybe verify the use case. I think the other bug was with snaps too [08:18] Is there any snap containing an Xwayland binary that might get run? [08:18] I woulnd't have thought so [08:22] Morning oSoMoN [08:22] duflu, I don't think so [08:23] duflu, yeah, all classic snaps it seems [08:24] willcooke, I wonder if they're special because they're over-confined, lacking access to the graphics hardware, and so hitting the swrast use case [08:25] A crash in Xwayland is still a crash in Xwayland, but I'm trying to describe the trigger [08:25] hey duflu [08:27] duflu, well, if they're classic snaps they shouldn't be confined at all [08:27] oh, that's what you're saying, they are being confined when they shouldnt be [08:28] I'm not really sure. I'm not looking at it at the moment [08:28] thanks duflu [08:28] jibel, I have a present for your from the OEM team :) [08:29] The good news is people are still fixing Xwayland, including some unreleased fixes [08:32] willcooke, oh, should I be pleased or not? ;) [08:32] it's a giant 💩 [08:32] jibel, not sure yet. It's a 15" 2-in-1. Quite a nice machine. [08:32] just on time for the release [08:32] Laney, I'm ok, that cold is still annoying me though, I didn't sleep well and now it seems the baby got it as well so didn't sleep well either which over-screwed my night :/ and he had fewer this morning so isn't at the childcare so we are looking after him in turn today (well he's sleep for now so we are both working atm ;) [08:33] jibel, I took Bionic for a spin on it, it's fine. Everything works [08:33] seb128, I found the problem with the screenreader, I'm testing a fix [08:33] it's caused by https://github.com/GNOME/gnome-shell/commit/01d88119132f277196fa39c05b733dd1b4c5e951#diff-27bd3da1798473ae0fa876e6712c08a4 [08:34] jibel, nice one! [08:35] jibel, that sounds weird/doesn't match my understanding of the issue [08:36] jibel, is the issue you are looking at happening in a live session/after installation or in ubiquity only mode? [08:36] seb128, in maybe-ubiquity mode [08:38] seb128, when you're on the "try or install" window the a11y icon is not displayed and the screen reader cannot be activated with alt+super+s [08:39] jibel, what is "maybe-ubiquity"? [08:39] seb128: hugs, hope everyone gets better soon :( [08:39] the first screen that give you the choice between live session and install only? [08:39] Laney, thx [08:39] maybe-ubiquity is the try or install screen [08:39] thx [08:39] is the issue specific to that screen? [08:40] seb128, the ubiquity-dm window with a list of language and 2 buttons "Try Ubuntu" / "Install Ubuntu" [08:40] like does it work then in live session or install modes? [08:41] jibel, anyway, I think that's not he issue [08:41] or maybe you would get the a11y icon to work [08:41] but that still wouldn't work to enable the screenreader [08:44] I poked a bit a it and remembered I debugged it in N-Y, it's still the same issue than by then [08:44] like me try to find the bug, I had a start of patch that needed finished iirc [08:46] ah, found it [08:46] jibel, it's still https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1719995 [08:46] Ubuntu bug 1719995 in ubiquity (Ubuntu) "Screen reader does not read Ubiquity windows" [High,Confirmed] [08:46] basically what the keybind and shell menu do is to set the screenreader enable key in gsettings to true [08:47] which in the session enable the service through gnome-session/autostart conditionned on this key value [08:47] but we don't use gnome-session in maybe-ubiquity or ubiquity-dm [08:47] so nothing watch the gsettings key to activate the service when it's true [08:48] my start of patch was https://launchpadlibrarian.net/340645974/ubiquity-screen-reader.patch [08:48] but it would at least require stopping the service if the keybinding is used again [08:48] jibel, ^ can you test that patch though? [08:49] I'm unsure how to patch things from casper-bottom [08:49] or if that needs a local iso respin [08:49] so I didn't test it [08:49] but in theory it should work [08:50] didrocks, ^ btw, you were interested by that screenreader issue as well [08:58] seb128, I'll test it [09:01] jibel, thx [09:01] seb128, you meant orca instead of onboard? [09:01] in the patch [09:01] jibel, yeah, sorry [09:01] I keep confusing those 2 :/ [09:01] np [09:02] there is ton of dead code in casper related to a11y [09:02] jibel, the UI side/lack of icon is another problem and that might have to do with the commit you pointed, I don't know [09:03] there is a gnome-shell ubiquity mode that restrict the UI [09:03] so it's probably a matter of making the a11y icon displayed in that mode [09:03] seb128, yeah, the component a11yGreeter has been dropped and now it's only a11y + setting always-show-universal-access [09:04] so it's just changing the name in the ubiquity mode and setting the key in a casper script [09:05] good [09:05] also there is a key for /o/g/settings-daemon/plugins/media-keys/screenreader set to '' [09:08] where? [09:10] seb128, when you dump the registry in ubiquity-dm mode. This key was not present in previous releases [09:11] not set [09:14] Laney, hi [09:16] jibel, that sounds weird [09:16] Laney, did you see my messages about the gs/pstotext problem yesterday night? [09:22] tkamppeter: yeah, I think it's OK to switch the tests to use ps2txt if you want - in that case it should be filed in Debian on doc-rfc too IMO [09:27] jibel, do you know why the pending isos didn't get promoted to current for some days? [09:31] seb128, I'll have a look [09:33] jibel, thx [09:34] seb128: good work :) [09:34] didrocks, thx [09:57] Laney, then I suggest to file it in both dhelp and doc-rfc in Debian, to not only switch over the test but assure that dhelp stays functional and to completely eliminatre the 14-year-untouched pstotext from the distro, removing a dependency, removing an upstream-unmaintained package, removing a package which does a task which ghostscript does by itself with ps2txt. [09:58] tkamppeter: well, I filed an RC bug so if pstotext's maintainer wants to fix it they can and if they don't it'll be auto-removed, but that shouldn't stop other packages from moving away [10:07] willcooke, it appears all my performance problems have been caused by changing hardware Haswell->Kaby Lake. So no code regressed, I just never encountered the problem when working on Haswell. Now with the same HDD in the old machine, performance is sweet again. Progress, but weird. [10:08] Oh, wow, odd. [10:08] My Haswell is 4x faster than the Kaby Lake [10:09] Is your Kaby lake dual core with hyper threading? vs 4 actual cores on Haswell? [10:09] willcooke, they are both 4 actual cores (desktops) [10:09] Although the Haswell is a Xeon so might be extra-fancy [10:27] seb128, I tried your patch, it actives the screen reader but then I hit bug 1719995 [10:27] bug 1719995 in ubiquity (Ubuntu) "Screen reader does not read Ubiquity windows" [High,Confirmed] https://launchpad.net/bugs/1719995 [10:27] activates* [10:28] seb128, can you submit an MP? [10:29] you have to import Gio too and fix the pep8/pyflakes issues [10:55] Laney, I have moved doc-rfc to ps2txt now and uploaded it, it got immediately approved (was it you?). The FTBFS is not caused by my change, it broke much earlier, in the phase of dependency installation, probably simply needs to be restarted. [10:57] It's already restarted so I can't see the error. [10:57] And no, there is a script to do it for unseeded-un-packagesetted packages. [10:57] Please could you send that on to Debian? [10:58] jibel, well, as written on the bug the reason I didn't submit it is because I didn't implement the desactivation, but I can submit the part I have sure [10:58] jibel, is the "does not read ubiquity windows" also true in the live session? [10:59] Laney, I have restarted the build after writing the last message to you. [10:59] Laney, to where to Debian should I send it? [11:00] the BTS [11:02] e.g. using 'submittodebian' from the source package you just uploaded [11:02] tkamppeter: Looks like it worked this time [11:26] kenvandine[m][m]: hey! When you are around: do you have a minute to discuss the theme snap? (There are multiple issues/things to coordonate upon) [11:35] Laney: jibel: this commit sounds interesting when ubiquity was really slow to log in https://gitlab.gnome.org/GNOME/gnome-settings-daemon/commit/89492d1d [11:37] jbicha: sounds like it could be [11:39] I don't think it was only gsd-media-keys failing though [11:40] is it in .1? [11:40] g-s-d didn't do a .1 yet, but it's at least in the gnome-3-28 branch [11:40] it's in the branch at least [11:41] ah yes [11:46] for those that didn't hear, garnacho is going to be Red Hat's g-c-c/g-s-d maintainer from now on [11:53] GNOME one as well? [11:54] Bastien mentioned yesterday that he was going to be the packages maintainers in their distros [11:54] but upstream was not clear to me [11:57] he said "sure" but it sounded a bit unexpected :P [12:01] my understanding was that was supposed to be part of the job description for the Red Hat job 🤷 [12:02] let's hope it doesn't mean he stop fixing input bugs or issues with the new OSK because he's too busy with other things now [12:03] so um we managed to get the OSK extended characters working in F28, not sure what's not working right in Debian/Ubuntu though === pstolowski is now known as pstolowski|lunch [12:48] kenvandine: not sure if you saw my ping as the one I had was against your matrix instance (regarding gtk-icon-theme) [12:48] hey didrocks [12:49] kenvandine: hey hey ;) it was for reference: 13:26:48 didrocks | kenvandine[m][m]: hey! When you are around: do you have a minute to discuss the theme snap? (There are [12:49] | multiple issues/things to coordonate upon) [12:50] didrocks, it's definately not a perfect solution [12:50] kenvandine: yeah, and in particular that it's not building right now :p [12:51] but it was the best we could come up with after 3 or 4 different meetings in budapest [12:51] (not in the store either and no snap referencing it) [12:51] should we have a quick HO to see how we can progress on it? [12:51] it's on my list to discuss with james this week [12:51] i'm not sure of the current status [12:52] kenvandine: can we discuss about the additions I want to make for communitheme, sharing snap and such? [12:52] at least so that I can hook up on my side [12:52] Laney, pstotext-free doc-rfc and dhelp uploaded and Debian bugs reported. [12:52] we need to get communitheme included for sure [12:53] i'll have james add that [12:53] didrocks, can't do a HO right now [12:53] kenvandine: well, there are some subtilities for people to test latest :) [12:53] tkamppeter: thanks, what is pstotext-free? [12:53] and multiple points I want to ensure we get that correct [12:53] no headphones and i'm working from the car dealer :) [12:54] kenvandine: ahah, best work place ever! Will have time later? [12:54] yes [13:01] kenvandine picking up his tesla [13:01] Laney, i wish! [13:02] getting service done [13:03] seb128, the screen reader works in the live session but doesn't in "install" mode and ubiquity dm [13:03] even when orca is started? [13:03] :/ [13:04] but it reads other part of the UI like the indicator? [13:05] no it doesn't read anything [13:06] like orca is not started [13:06] well, the indicator tells it is [13:06] it is started [13:06] let me see if I find any error [13:07] does it tell "screen reading activated" when being started? [13:07] or maybe sound is not working? [13:08] the sound is working, the drum is playing, but when the screen reader is activated it does not tell "screen reader activated" [13:11] weird [13:11] well let me know if you can figure out something, the other day when I debugged and started orca manually it was reading fine [13:12] orca is not running [13:12] what's the easier way to test my ubiquity patch? is that to rebuild an iso? [13:12] hum [13:12] I'm lost [13:12] that's after testing my patch? [13:12] yes [13:12] but the patch didn't work/start orca? [13:12] and changing onboard by orca [13:13] seb128, the patch works because the indicator shows up and the screenreader is enabled but orca does not start [13:13] wth [13:13] the patch does subprocess.Popen() on it [13:14] so that call fail? [13:14] any error in the log? [13:14] that doesn't make any sense [13:14] what my patch does is basically react to the gsettings key status change and start orca [13:14] you can probably add systemd.unit=rescue.target to break in a more functional environment [13:15] don't think it has to be as early as casper-bottom for this [13:15] it shouldn't change the indicator behaviour [13:15] in there you can have network / vim / stuff [13:15] Laney, ah, good to know, thanks [13:15] you add emergency to the boot command line [13:15] there are many ways... [13:16] well maybe with https://launchpad.net/ubuntu/+source/ubiquity/18.04.5 on the pending iso I can just go to a VT now? [13:16] VTs are supposed to work again now too, so maybe you can even hack it live and restart the session (not tried that) [13:16] * seb128 is downloading the pending iso [13:17] I hope it works, it didn't migrate to current [13:17] unsure if that's because it's buggy :p [13:17] seb128, yes use the pending iso [13:17] you have VTs [13:17] good [13:17] * seb128 tries that [13:17] and it works [13:18] seb128, the promotion failed because the server running the test cannot download the iso [13:22] jibel, thx for checking === pstolowski|lunch is now known as pstolowski [13:30] seb128, starting orca from a console works so it looks like the callback is not called [13:31] right [13:31] that makes more sense [13:31] jibel, what is "emergency" on the boot cmd supposed to do? [13:33] give you a shell [13:34] I'm not really sure why he said that since I already gave a way to do it, it's just confusing to give another one [13:34] and they aren't equivalent, you don't get as many services started in emergency.target vs rescue.target [13:34] it got late, I should have lunch [13:35] 🍽 [13:35] hmm, I wonder if the Characters search provider should prefer emoji over other things [13:36] I searched for fork and didn't get 🍴 [13:37] kenvandine: how did my Characters snap get disconnected from the gnome-3-26-1604 snap? [13:39] $ snap changes error: no changes found [13:41] anyway: https://forum.snapcraft.io/t/gnome-shell-search-providers/4322 :( [13:42] jbicha, ugh... don't know [13:43] must be a snapd bug [13:43] where would I even look for logs for stuff like that? [13:43] jbicha, we need to spend time on figuring out the search providers [13:43] i really don't know [13:43] popey, any idea where we can get logs for snaps disconnecting? [13:43] maybe the journal [13:43] I under re: the search providers. L_aney just made me jealous :) [13:43] #snappy might be a better channel [13:44] hah [13:44] good point [13:44] I accidentally revealed how I have it installed [13:45] Laney, pstotext-free is free of pstotext, without pstotext. [13:45] understand [13:46] Laney, doc-rfc has made it into Bionic now. [14:16] kenvandine: i dont understand [14:18] popey, wasn't sure where to look for logs on why interfaces might not be connected that should have already been autoconnected [14:19] for jbicha, gnome-characters wasn't connected to gnome-3-26-1604 [14:19] but turns out after a reboot it was [14:19] so not sure why it thought it wasn't connected [14:19] we sent him to #snappy but i don't think anyone responded to him [14:23] huh [14:23] Laney, sorry, I went with trying the emergency tip first because I was unsure how systemd.unit work, if that's a boot option or a target I need to create on the fs i some way [14:30] seb128, do you understand why with your patch the a11y icons shows up when the screenreader is activated and why it doesn't without? [14:30] icon* [14:36] jibel, no, that doesn't make sense to me [14:37] jibel: are you talking about the gnome-shell accessibility menu in the top right corner? [14:37] you are sure it's not another update on a recent iso that made a difference or something? [14:37] seb128, to me neither and ij both cases the value of the key changes [14:37] jbicha, yes [14:37] seb128, I'll recreate the iso to be sure [14:38] you test by recreating the iso each time? [14:38] yes and no, it depends. [14:38] jibel: gsettings set org.gnome.desktop.a11y always-show-universal-access-status true [14:38] using the "emergency" boot option I could wget the patch and apply it on the iso [14:39] jbicha, nothing does that in ubiquity atm though? [14:39] jbicha, I know but the problem is that orca is not starting [14:39] seb128, this is the patch i pointed out this morning [14:39] GNOME3 initially always had that menu showing but people complained that they didn't need universal access stuff [14:40] they removed a11yGreeter but now you have to set always----status = true [14:41] I think it makes sense to always show the indicator on the live session [14:41] at least for installation [14:42] jibel: yes at least for the install mode where you aren't running a full desktop session [14:42] on the other hand, I guess we'd need to make sure all those toggles actually work in that environment [14:43] but having the indicator *and* orca is even more useful :) [14:45] +- [14:46] ups [14:46] jbicha, right, they don't atm which is what we are talking about [15:00] tkamppeter: oh right, I get it, I thought that was a package name [15:00] seb128: it's a kernel cmdline parameter the same as emergency, it tells systemd to start a different unit [15:01] instead of? [15:01] default.target [15:01] or just adding an unit? [15:01] ah ok [15:07] hello [15:08] hi Trevinho [15:08] good morning Trevinho [15:08] hey Trevinho [15:12] Laney, noe only poppler and dhelp need to be made using the autopkgtest of the new doc-rfc, which they seem not to use according to http://people.canonical.com/~ubuntu-archive/proposed-migration/bionic/update_excuses.html. [15:12] Laney, s/noe/now/ [15:14] tkamppeter: They just need retrying. [15:14] I did it [15:16] we've got a few FTBFS on the rebuild test btw http://qa.ubuntuwire.org/ftbfs/rebuilds/test-rebuild-20180408-bionic.html [15:16] in case anyone wants releasey tasks to do [15:26] hi didrocks, oSoMoN and seb128 :) [15:27] and Laney : [15:27] oh hey Trevinho [15:27] what's up? [15:27] all good, at the end here the network can be better than expected if you go in the right spots and know the right people :-D [15:28] Raúl your big buddy [15:30] ahah, sort of [15:31] Hi Laney, time to look at the dh_translations patch? [15:31] GunnarHj: not right now [15:31] Laney: Ok. [15:35] jibel, Laney, jbicha, the patch doesn't work because ubiquity-dm is root so it listens to the gsettings key for that user where the keybinding is going through gsd-media-key which is an "ubuntu" user process [15:37] I wish we still had upstart :/ [15:37] it would be trivial to do an user job use the gsettings key as a condition [15:39] seb128: oh right, do you need to do the drop_privileges thing? [15:39] (I didn't see your patch) [15:39] I guess [15:40] we had to do something like that for the gsd-xsettings thing [15:40] that's waiting for a dbus signal so not exactly the same [15:40] but could be comparable [15:40] because you had to get on the user's session bus [15:41] Laney, ubiquity-dm does a bunch of those, gtk_ui not atm it seems? [15:41] Laney, my patch is people.canonical.com/~seb128/ubiquity-screen-reader.patch [15:41] http://people.canonical.com/~seb128/ubiquity-screen-reader.patch [15:41] huh, that was for usd only [15:41] hope we don't have that bug back since it's not being used any more [15:43] seb128: you can use settings_a11y.connect("changed::screen-reader-enabled", on_a11y_changed) [15:44] and maybe also gsettings.get_boolean() == True [15:44] GET_BOOLEAN EQUALS TRUE [15:44] GET OUT! [15:44] != False [15:44] :D [15:44] :) [15:44] Laney: Btw, I just realized something which I ought to change in that patch. Please wait until I have submitted next version. [15:44] Trevinho, that would work for any user? [15:45] Trevinho, or what do you mean? [15:45] seb128: look at the slideshow bit (start_slideshow()) [15:45] no. no.. just cleanups :) [15:45] it drops, does a load of stuff to set up webkit and then regains [15:45] Trevinho, ah [15:46] Laney, right, I see now, going to try that, thanks [15:48] instead of dropping the privileges you might just set the proper env to connect to user gsettings no? Like the runtime path and dbus session [15:48] well, for monitoring changes just the session should be enough [15:49] dbus will refuse that I think [15:49] plus that is gruesome [15:49] and we have handy functions to drop/restore privileges [15:49] so better to use than invent another solution [15:50] andyrock was using that I think too, so maybe he has tips [15:50] yay poppler went in [15:50] thanks tkamppeter [15:50] I was using what? :) [15:52] andyrock, ubiquity's misc.drop_privileges_save() [16:11] was not working properly in my case [16:11] *it was [16:11] I don't remember how [16:11] I had to use gtkplug to do the privilege drop properly [16:12] iirc because I could not drop the privileges at that point [16:12] andyrock: the livepatch page in gnome-initial-setup doesn't work very well :( I guess I'll file another bug [16:12] jbicha: what's the problem? [16:13] Laney, thanks, so I have hit Retry a little bit too early. Nice that all is working now. pstotext has no reverse dependencies any more, so it is ready to be shot to the moon. [16:13] andyrock: have you tried it? [16:13] gnome-initial-setup 3.28.0-2ubuntu3 was just published to bionic [16:13] jbicha: please make sure you remove your u1 account from gnome-control-center === pstolowski is now known as pstolowski|afk [16:13] otherwise it won't work [16:17] tkamppeter: next time wait until rmadison shows you the required version is available [16:17] bit of a waste of compute cycles to spuriously rery [16:17] retry [16:17] if there's no resolution to the RC bug I filed then it should hopefully be removed from unstable - we could follow that [16:17] andyrock: oh, it worked this time [16:18] maybe I was hit by the "you can't just close Online Accounts immediately after removing an account" bug you fixed in g-c-c 3.28.1 [16:19] mmm that should be fixed [16:19] not in bionic [16:19] upstream at least it's fixed [16:19] yes [16:19] Laney, OK. [16:19] jbicha: so I can make it working if you already have that account [16:20] jbicha: one possible solution is e.g. if we get the error "account-already-setup" [16:20] andyrock: I like the way it works in software-properties [16:20] we can just pick that account [16:20] jbicha: https://github.com/CanonicalLtd/desktop-design/issues/62 [16:20] CanonicalLtd bug 62 in desktop-design "Authentication in gnome-software using gnome-online-accounts" (comments: 9) [Priority: High, Open] - Assigned to matthewpaulthomas (Matthew Thomas) [16:21] bug 62 in Launchpad itself "Maintainers can add themselves to the CC: list too" [Medium,Invalid] https://launchpad.net/bugs/62 [16:21] hmm, there are some interesting papercut bugs here [16:21] jbicha: I'm going to change the design once it is completed [16:22] then we can use the same design in software-propertis, gnome-initial-setup and gnome-software [16:22] but we really need to work with goa upstream to move that dialog in their API [16:24] andyrock: for instance: https://bicha.net/i/software-properties-livepatch-half-on.png (that's what it looks like after completing g-initial-setup) [16:24] all I have to do is click Sign In and choose my account in the little popup dialog and it then displays correctly [16:26] oh I can fix this [16:26] Can you open a bug? [16:27] sure [16:39] jbicha: thx! [17:00] G'night all. I'm making pizza from scratch for dinner. Sorry Trevinho and andyrock, I think you will be horrified. Pics to follow if it works. [17:02] k, so dropping the ubiquity priviledges as in people.canonical.com/~seb128/ubiquity-screen-reader.patch doesn't work [17:02]