=== arraybolt3_ is now known as arraybolt3 [02:27] RikMills, arraybolt3: regression in plasma-workspace broke Ubuntu Studio's SDDM theme in 22.04: https://bugs.launchpad.net/bugs/2003388 [02:27] -ubottu:#ubuntustudio-devel- Launchpad bug 2003388 in plasma-workspace (Ubuntu Jammy) "sddm-theme-breeze breaks Ubuntu Studio login" [Undecided, New] [02:28] Eickmeyer: Noticed that, pitched in on the bug report. [02:28] Didn't know it was plasma-workspace, looking. [02:28] Yeah, but the askubuntu question is entirely incorrect. [02:29] Site is in read-only mode, can't do anything now. [02:29] Yeah, until Midnight Eastern, I assume. [02:29] I'll help with it once it gets untangled. [02:30] I love how people assume it's ubuntu studio's fault without investigating to see whether or not it's a symlink, which this is. [02:30] lol well sometimes people make mistakes. FWIW, though, I'm not "Organic Marble [02:30] Oh, I know. [02:31] * arraybolt3 cracks knuckles and reviews recent changes to plasma-workspace [02:32] * arraybolt3 spins up VM [02:32] (Nothing interesting was in the changelog, sadly.) [02:33] I remember I had to resolve this with Kinetic, but that was merely a workaround where I copied the qml file and removed the references to the missing properties. [02:33] Lunar uses an entirely new theme. [02:33] I remember this also. It was that bug I failed to report until the absolute last minute :P [02:34] I could do the same for Jammy, but I'd rather not because that's not fixing the problem, that's just working around it. [02:34] I guess it depends on what the problem exactly is and how hard it is to really fix. If it's just a botched config file or something, no problem. If it's an upstream KDE bug... [02:36] Right. In this case, it wasn't a botched config file. It was an upstream KDE bug that I just created a workaround for. Ugly, but it worked. [02:36] Going forward, though, Studio has an entirely new SDDM theme that comes from materia-kde. [02:38] Personally, I'd just use the workaround and submit the real bugfix to upstream KDE, then backport it once it's accepted. Otherwise we have vendor-specific patches (I think that's what they're called) in Ubuntu that aren't in KDE and that seems like a bad thing. [02:39] My question is, since it's the same theme that's used by Kubuntu just with a different wallpaper, why isn't Kubuntu affected? [02:39] Maybe they are? [02:40] I mean it's possible they've got the same bug and no one's reported it. I noticed this bug earlier (like, yesterday or the day before) and didn't say anything since I wasn't sure if it was just on my side or not. [02:40] I probably would have reported it if it persisted even after having everything upgraded. [02:40] I guess I may as well get my Kubuntu ISO copied over so I can make a VM with it too. [02:41] (For once I managed to keep all my ISOs across an OS reinstall \o/) [02:41] Nice [02:46] Alright, I now have an Ubuntu Studio VM updating and a Kubuntu VM installing. [03:35] arraybolt3: Keep me posted, I'm probably checking-out for the night. I've got some sort of illness going on. [03:38] Will do. Sorry to hear you're under the weather. [04:26] I have confirmed that this problem does not appear to affect Kubuntu. [04:31] I can see right off the bat that there are differences between the Main.qml file in Ubuntu Studio and the one in Kubuntu. [04:31] Neither VM has relevant packages phasing. [04:33] And Kubuntu's file does not have the offending "actionItemsVisible" line in its Main.qml file. [04:35] There are some very significant differences between the files. The diff: [04:35] * arraybolt3[m] sent a code block: https://libera.ems.host/_matrix/media/v3/download/libera.chat/dc622d38928556bbaf5bdd3b50441bcc8eba258e [04:35] (The "KUBUNTU" and "STUDIO" lines in the comment at the top were added by me so I could tell them apart easily.) [04:36] arraybolt3[m]: How is that possible when one is a symlink to the other? [04:40] Well... [04:40] I'm not making this up. '''usr/share/sddm/themes/ubuntustudio/Main.qml -> ../breeze/Main.qml''' [04:40] *shrug* [04:40] * arraybolt3 digs deeper [04:41] On my Jammy VM, Main.qml is **not** a symlink. [04:41] Eickmeyer: I assume you're running Kinetic? [04:41] Because on Jammy, the Main.qml file is just a file. [04:41] I'm looking at the sources. [04:42] * arraybolt3[m] sent a code block: https://libera.ems.host/_matrix/media/v3/download/libera.chat/20b7cda5fcb963ec9445d207f062d67e904eda5d [04:42] That's what "ls -l /usr/share/sddm/themes/ubuntustudio" shows me. [04:43] https://git.launchpad.net/ubuntustudio-default-settings/tree/usr/share/sddm/themes/ubuntustudio/Main.qml?h=jammy&id=cef1f265372b5d5bc758f6b97f2bbc4d635dd652 [04:43] The problem is in Jammy. [04:44] * arraybolt3 does a pull-lp-source to see what's going on here [04:44] Not in Kinetic. [04:44] Right, I thought you were looking at files on your system. [04:44] No, I'm on Lunar. [04:44] I thought maybe the file was a symlink in Kinetic, not in Jammy. So nvm. [04:44] I'm looking at the sources. [04:44] I just ran "sl" rather than "ls" XD [04:44] hehe [04:45] When I do "pull-lp-source ubuntustudio-default-settings jammy", the Main.qml file is not a symlink. [04:45] The exact version is "ubuntustudio-default-settings-22.04.26.1" [04:45] * arraybolt3[m] sent a code block: https://libera.ems.host/_matrix/media/v3/download/libera.chat/122c5972b748b4e3a0163fe3210ca933c4e84d9d [04:46] "ls -l" output again. [04:47] Soooo... I'm guessing the file wasn't a symlink at some point, and now in Git it is, but the change that makes it a symlink hasn't been updated? [04:47] s/updated/uploaded/ [04:47] I just did the pull myself, which is very odd. [04:49] arraybolt3: Something got crossed, because looking at the debdiff, it was a symlink prior: https://launchpadlibrarian.net/623379853/ubuntustudio-default-settings_22.04.26_22.04.26.1.diff.gz [04:49] Very odd behavior. [04:50] I'm not sure if that's the link you meant to send, since I see nothing about Main.qml in there. Maybe I'm just missing something. [04:50] But I guess we just diff the git version with the archive version and get them to match? [04:51] Also, I thought you were going to bed :P [04:52] arraybolt3: I thought I was too. [04:53] So, what I'm seeing here is that the version that was uploaded matches what I'm seeing in the git history. The git history shows it was always a symlink for the past 3 years until just recently when I made the new sddm theme. [04:54] What was uploaded, and the debdiff (https://launchpadlibrarian.net/623379853/ubuntustudio-default-settings_22.04.26_22.04.26.1.diff.gz) shows that. [04:54] But, what pull-lp-source, and the package that gets downloaded, shows something completely different. [04:55] Very, very odd behavior on Launchpad's part. [04:55] Weird. [04:56] Then where in Canonical's crazy infrastructure did it find the file it's using now?! [04:56] Perhaps somehow the file from Kinetic managed to "leak" into Jammy? Or maybe even the one from Lunar? [04:56] That file most likely matches the version in Kinetic. [04:57] Either way, something leaked [04:57] It might be that it's not showing in the debdiff because maybe the upload decided to upload a whole entire tarball and not just a debdiff? [04:57] I don't even know if that's possible though. [04:57] It's not possible. The git history doesn't show it. [04:58] Worth a bug report against Launchpad itself? Or do we just say "well that was weird" and patch it? [04:58] I made sure to branch it early enough. [04:59] I wonder if maybe somehow an upload intended for Kinetic accidentally became a Jammy upload. [04:59] The debdiff would show that. [04:59] Yeah, you're right. [05:00] Where I'm at is I'm going to have to grab the Main.qml from Kinetic and throw it in Jammy because it fixes the exact error being reported. [05:00] Or maybe there's some way to get the symlink to actually go where it's supposed to? [05:01] No, I think the Main.qml version somehow leaked from an old version of sddm-theme-breeze now that I look at the diff better. [05:01] Better just to get what I know works. [05:01] Well, whatever works. It won't do any harm for it to not be a symlink. [05:02] Exactly. Even if it's copied from the breeze theme, no harm no foul, but I fear the functions were removed from frameworks, which is why the errors are happening. [05:03] What's odd is that the versions in Ubuntu Studio's Main.qml are *newer* than the ones in Kubuntu, I think? [05:03] But whatever. Further investigation is probably good for nothing but "huh, that's so weird" value. We've figured it out and have a fix planned. [05:03] We can speculate all we want, but what I know is that something got crossed somewhere and now we have to fix it. Problem for tomorrow's Erich. [05:03] +1 [05:04] * arraybolt3 shuts down VMs and keeps them around for future diagnostic purposes if needed [05:04] Thanks, I appreciate it. I'll probably open a bug report against Launchpad itself tomorrow, because this is just weird. [05:05] :) Glad to be able to help with something. [05:05] You're always a help with something. [05:05] (I may need to do some VM cleanup, I have, I kid you not, 15 VMs between Boxes and VirtualBox.) [05:06] Something tells me that having 16 operating systems installed on the same computer is a bit much. [05:07] Though then again, they're not taking up that much room. Nevermind. [05:07] I've heard of worse. [12:11] muon removed from debian unstable at least [14:47] https://bugs.launchpad.net/ubuntu/+source/plasma-workspace/+bug/2003388/comments/11 [14:47] -ubottu:#ubuntustudio-devel- Launchpad bug 2003388 in plasma-workspace (Ubuntu) "sddm-theme-breeze breaks Ubuntu Studio login" [Undecided, Opinion] [17:35] :( I liked muon. [17:37] Bridge to IRC seems broken. [17:37] !ping [17:37] pong! [17:38] lol now it works. [17:55] arraybolt3[m]: So, apparently some wires got crossed with the upload (I inspected the tarball), but it didn't show up in the debdiff. I'm flabbergasted. [18:10] :-/ [18:11] Well, hey, it's better than if Launchpad outright modified the uploaded tarball. That would have broken signing, right? [18:20] in the traditional sense yeah [18:21] but some stuff is weird sometimes ;) [18:26] Just weird how the incluseion of *a whole new file* didn't show up in the debdiff. That's what's tripping me up. [18:26] *inclusion [18:39] Unrelated: I suppose it might be unethical to launch the person into orbit that thought it would be a good idea to make pipewire-alsa a recommends of pipewire-pulse causing our images to ftb. [18:39] * Eickmeyer blacklists pipewire-alsa [18:40] We still need the pulseaudio package as several of our packages still have a hard depends on it. [18:47] http://launchpadlibrarian.net/647075684/ubuntustudio-default-settings_22.04.26.1_22.04.26.2.diff.gz [18:47] that debdiff shows no file change either! [18:52] RikMills: Yeah, I can't imagine it will, but if you inspect the tarball, this Main.qml has no contents because it's a symlink. [18:53] Like I said in the SRU template, I'm wary of the debdiffs right now. [18:54] Moral of the story: don't always trust debdiffs. [18:54] Which is sad, because that's what we've been trained is trustworthy. [18:56] I wonder if debdiffs even pick up symlinks? I know they don't handle permissions. [18:56] Like maybe they're not actually able to detect when a symlink gets... I mean idk, it still just seems weird and broken. [18:56] They should show every single file difference. [18:57] So, a symlink to an actual file should show up. [18:57] And the reverse should be true as well. [19:00] maybe if debdiff sees that a file is specified in debian/something.links so a symlink, it skips adding a diff for it? [19:00] probably not, but there has to be some fail in logic [19:01] * RikMills shrugs [19:13] Eickmeyer: LP weirdness continues. ubuntustudio-default-settings appears to have migrated to -proposed in Jammy... yet there are *no* binary files built from source. At least, not on the main package page, but then you click on the amd64 build and there's the files. [19:13] "sudo apt update" isn't detecting the new package (despite having enabled -proposed), even if I change mirror to the core one (archive.ubuntu.com). [19:13] arraybolt3: It takes a while to propagate to the mirros. [19:14] Right but I'm not using a mirror, right? Or is archive.ubuntu.com a mirror itself? [19:14] *mirors [19:14] archive.ubuntu.com is a mirror itself. [19:14] Oh. OK. [19:15] You never, from the outside of Canonical's infra, have direct access to the main internal server. [19:15] That makes sense. I thought for some reason that archive.ubuntu.com was the 'heart' machine that all others mirrored from, and that the ftpmaster.internal machine or whatever was like another "way" into that machine that was faster or something. I see I was mistaken. [19:16] *that all others ultimately derived from (lol if all mirrors drew from it that would be painful) [19:18] Yeah, that would be a huge load for one server. [19:18] I guess while I wait for that to find its way to a more easily accessible server, I'll start work on an XScreenSaver regression SRU. [19:18] (XScreenSaver is currently badly busted in Kinetic for everything, and in Lunar for everything but Lubuntu.) [19:18] Yeah, I'm working on this pipewire-alsa - pulseaudio conflict issue. I'm going to have to ask for help with it. [19:18] *in Lunar for everything (I think) but Lubuntu. [19:22] Eickmeyer: I thought blacklisting only made the build fail if a package got in, not prevented it from getting in? [19:22] (re your last query in -release) [19:22] It should prevent it from getting in. [19:22] If it's a recommends, it's not required, hence it should be blacklistable. [19:23] Oh, because it's a *recommends*, I get it. [20:07] arraybolt3: The mirrors are being molasses in this month slow. [20:07] Yeah, noticed that :P [20:07] At least the bug reporter got it verified. [20:08] Just tried another apt update, no dice. [20:08] Should I just go ahead and do a manual download from LP? I feel like that's not quite the right way to verify things. [20:24] arraybolt3: Just checked, the mirror is good nw. [20:24] *nwo [20:24] *now [20:24] Verified myself, it looks good. [20:25] But more eyes the better [20:25] Downloading and testing. [20:33] Mind if I mark the bug as verification-done? [20:33] Eickmeyer: ^ [20:33] arraybolt3: Please do! [20:34] arraybolt3: Then ping vorlon in -release [20:35] Done. [20:36] Thanks! [20:50] foo-1.5.3.1532.34775+really1.5.2.123.4346-7+dfsg3ubuntu1.22.04.1 ftw! [20:51] (That's actually a valid version number, I think. And I still didn't add an epoch or a bpo.) [21:23] It is a valid version number, but still warrants an ew.