=== guiverc2 is now known as guvierc === guvierc is now known as guiverc [00:20] [telegram] https://matterbridge.lubuntu.me/a861b88f/file_10228.jpg [00:23] [matrix] That's pretty cool too [00:25] hmm, the window will end up covering the hummingbird [00:25] (assuming we "desktopify" it) [00:32] tsimonq2: I'm slightly unsure about making the theme of the button match the theme of the ISO welcome screen, and am thinking perhaps we should do the reverse. Rationale - rounded corners are cool and all, but they're virtually absent from the LXQt desktop and only exist on the website and the welcome screen. It's a lack of consistency that *can't* [00:32] be fixed in LXQt, whereas it's relatively trivial to just use sharp corners in the welcome screen and website. IMO that would look better than the alternative. Thoughts? [00:33] (er, correction - it's almost certainly way harder to make LXQt use rounded corners than it is to use sharp corners elsewhere. I also kinda like sharp corners :P but that's probably just me) [01:10] ITP bugs filed for picom-conf and sddm-conf in Debian. [01:12] [telegram] Thank you! [01:15] * arraybolt3 checks The List to see what pressing things need done... [01:15] ...probably either Wayland experimentation or fixing the bug in our start screen that can lead to interrupted installs [01:28] guiverc: just got a grub update screen during installing updates using the Lubuntu upgrader (not using apt directly). Even though I chose a disk to install GRUB onto, I was told I had picked no disks. Have you seen this recently? [01:29] * arraybolt3 wishes I had stopped and gotten a screenshot, oh well, I can probably reproduce it [01:29] sorry I cannot picture what you're asking about; so almost certainly no. [01:31] try pressing "Install Full Upgrade" on a Noble machine you've not updated in a while if you have one of those [01:31] * arraybolt3 has an outdated ISO I can use otherwise, also this is probably specific to BIOS machines [01:53] tsimonq2: how would you feel about the actual installer prompt UI only being displayed on the largest resolution screen found, and just a background image being shown on the other screens? I think that would elegantly get around the problem guiverc noticed with being able to accidentally terminate an installation. [01:53] * guiverc loves that solution arraybolt3 [01:57] * arraybolt3 tries to implement it [02:05] heh, /me just learned that the startup window is *not* fullscreen, but rather is set to be just big enough that it appears to be fullscreen and the titlebar is hiding just outside the screen itself [02:05] ok so I'm going to do some cleanup work since that seems distinctly wrong to me [02:21] that ended up being pretty easy, time to test in VBox [02:34] sigh, just discovered a theming issue in the installer prompt :-/ [02:34] I'm guessing the symlink-fu I had in the CMake file at one point probably got either lost or isn't working right [03:00] guiverc: https://i.imgur.com/vYLlq19.png [03:00] better? [03:01] (code changes are already pushed) [03:01] yep looks good arraybolt3 [03:02] nifty, I'll leave a note on the bug report that the upstream code is fixed, then let tsimonq2 make a release once it's ready. [03:02] I do need to fix the symlink stuff too, so I'll try and see what's going on there while I'm right here. [03:02] * guiverc did like seeing that rat-possum like numbat creature though... [03:02] also I couldn't reproduce the weird GRUB configuration issue, so I guess that'll be a mystery lost to the sands of time [03:03] the numbat's still there, you just have to click "Try Lubuntu" to see it :) [03:04] :) [03:30] [telegram] guiverc: You do much more testing than I do these days, what would be the chances of someone having e.g. a broken laptop screen and an external monitor, having the screen go to the wrong place? [03:30] [telegram] It's an edge case, I'll admit; we should discuss how prevalent that edge case actually is. [03:31] hmm, the symlink that gets made by cmake to do theming right isn't getting made in the lubuntu-installer-prompt package, thus theming is a bit broken [03:31] [telegram] (Small button on the bottom saying "make this the active display" and only connect signals and such if the given screen is being used?) [03:32] [telegram] hmmmmm [03:32] [telegram] Time for a git bisect XD [03:33] it's not in the Git repo [03:33] the upstream code is good, I just tested it [03:33] it's the packaging's fault [03:33] I like the idea of a "make this the active screen" button, I might try to implement that soon-ish [03:33] tsimonq2, i used to test using a laptop with near-dead (unreadable) screen & external monitor... but most devices have FN key combinations that let you enable/disable/combination of screens which is what I'd use for problematic device.. ie. make machine show same output on both screens (mirrored) but OS think it's only got a single screen attached; meaning no coding is required for base OS [03:36] I wonder if that actually works in a pure Openbox environment before LXQt is started? [03:36] (by FN key; on one laptop there is a screen pic on F7 key which means FN+F7 and device flips between 5 options of internal-only, ext-only, int+ext-mirror; side-by-side etc... Fn-f7 on another ... ie. firmware of box offers option) [03:36] And I don't think all machines have that kind of key. [03:37] No they don't but most enterprise have it in my experience.. cheaper laptops may not.. [03:38] It's definitely worth investigating and supporting, and I have an implementation idea to do what Simon suggested a bit ago. [03:38] But at any rate, I should probably EOD now or pretty soon. [03:43] [telegram] arraybolt3: I think we're missing the runtime dependency for the extra widgets [03:44] Extra widgets? [03:44] The theming issue is because xdg-Lubuntu isn't getting symljnked to xdg-Lubuntu-Live-Environment (or whatever I set the CMake file to do, you get what I mean). [03:45] CMake makes the symlink at install time as it should, but the packaging isn't picking it up. [03:46] I realized the fix I just did isn't quite complete - what if the user only has one screen, but somehow alt+tabs their way back to the installer prompt (getting around fullscreen Calamares), then changes locale? Worse, what if we get the GUI for pkgselect fixed, then go back to running Calamares in windowed mode? [03:47] [telegram] We might be talking about two different things: KLed is a KExtraWidget [03:47] So we *really* need to hide or disable the locale changing box entirely when the user clicks Install. [03:48] tsimonq2: ah, indeed we are. I'm talking about weird theming in Cala, particularly the window titlebars that appear in popup windows. [03:48] [telegram] Sounds like a plan. We can use the existing WiFi show/hide functionality as an example [03:48] Also my most recent changes broke the installer prompt in a couple of ways for unknown reasons, so... I'mma go fix it I guess. [03:48] tsimonq2: don't make a new release yet please :) [03:49] [telegram] Did you ever consider it's because Calamares isn't running in an LXQt session and thus does not pick up our LXQt theme? Setting the appropriate XDG variables only happens when starting the user session, which is why you'll notice differences there [03:49] [telegram] I'm not doing any more work today ;) [03:49] That's exactly why it's happening, but my CMake file makes a symlink that fixed it. [03:50] [telegram] I mean... sure, that's one way of doing it [03:50] The name of the XDG session is "xdg-Lubuntu-Live-Environment" or something like that, so CMakeLists.txt symlinks /etc/xdg/xdg-Lubuntu to /etc/xdg/xdg-Lubuntu-Live-Environment. [03:50] And that symlink is getting lost at package build time even though CMake generates it. [03:50] [telegram] Ideally we'd literally move those XDG variable export lines up to the top of the Bash script and just call it a day :P [03:51] [telegram] When the installer prompt is called, it just acts like it's using xdg-Lubuntu [03:51] mhh, that might work but I seem to think we tried things other than symlinks without good results. [03:51] maybe not though, my memory is hazy [03:51] I'll try changing the Bash script while I'm right here and see what happens [03:51] [telegram] We didn't try this, I just inserted those myself last upload [03:51] [telegram] Literally move those export lines to the top of that file ;) [03:52] [telegram] Oh also [03:52] ah ok [03:52] yeah that surely will work then, since the LXQt session looks right if you choose to try Lubuntu [03:52] [telegram] That Calamares call in Install Lubuntu really needs to match the desktop entry exactly, however we do it, so we're not chasing bugs in functionality differences [03:53] [telegram] Between those two things, that *should* bring us a step forward towards better teeming [03:53] I like it [03:53] [telegram] *theming lmao [03:54] what in all creation [03:54] [telegram] Anyway, thanks! I'll tag in ~= the morning, or whenever you feel it's ready [03:54] 516 connect(calamares, static_cast(&QProcess::finished), [03:54] 517 this, [this](int, QProcess::ExitStatus){ this->tryLubuntu(); }); [03:54] who wrote that disaster [03:54] :P [03:54] [telegram] That would be me :P [03:54] [telegram] StackOverflow helped with that one... [03:54] I can't even figure out what that's supposed to pretend like it's doing [03:54] [telegram] If you can get a simpler version to work, go for it ;) [03:55] kk, I just found it since this is the part that's currently broken [03:56] [telegram] Here's the big thing: that QProcess must run asynchronously, or the entire installer prompt will freeze because it's running in the same thread [03:56] (just for the record I'm not trying to be mean, I was just making a joke because I was so confused trying to read it :P) [03:57] [telegram] No worries ;) [03:58] [telegram] You could move that QProcess to be a private variable belonging to InstallerPrompt and just connect it normally like the rest of the symbols, just make sure there's proper locking in place [03:58] [telegram] I keep thinking signals and typing symbols :P [04:02] connect(calamares, SIGNAL(finished(int)), this, SLOT(tryLubuntu()); // compiles [04:02] Sometimes old syntax is the best syntax [04:03] now let's see if that unbreaks it [04:14] hrm, it did not [04:15] for some reason if you try to cancel installation (thus closing Calamares), the installer prompt now locks up entirely. [04:15] which makes like zero sense to me because all that happens when you quit Calamares is it calls tryLubuntu(), which just calls QApplication::quit(). How can that lock up?! [04:15] (it locked up before the signal change too) [04:17] hmm, perhaps it's Calamares that is locking up [04:18] yeah because now it's just working, for some reason the first time I tried to cancel Cala it locked up rather than cancelling. I thought it was my code's fault :P [04:21] Final point of confusion - despite the fact that the env vars have been moved to the top of the script, and I changed the lubuntu-installer script to use "sudo -E calamares -D6" (which usually gives nice theming), the theming in Cala is still slightly off. Dunno what's wrong there, but I really should EOD now. I think I've not broken anything that [04:21] wasn't broken before, and the upstream code now has a fix for the critical bug guiverc reported, so I think that's good. [04:21] :q [04:21] oops, that was meant for my terminal [04:22] o/ [04:22] night arraybolt3 & thanks for your work [04:44] guiverc: thanks :) and thank you for noticing that bug, that could have caused a big mess otherwise [04:45] :) [06:02] labwc supports window snapping. The one feature I missed the most in Lubuntu may end up finally coming to fruition in 24.04. [06:03] that is like, really happy. That alone would make Wayland support worth it. [15:35] [telegram] https://bugs.launchpad.net/ubuntu/+source/calamares/+bug/2046500 [15:35] -ubottu:#lubuntu-devel- Launchpad bug 2046500 in calamares (Ubuntu) "Noble daily calamares installer does not honor existing partition table on 'Erase All'" [High, In Progress] [15:36] [telegram] Must be head down on dayjob for most of today tbh [15:37] [telegram] I am confused by that bug. Erase disk erases the disk. That wasn't the intention? [16:04] [telegram] I agreee with you Dan ..the report also mentions semantics - so maybe a misunderstanding ..also the installer always shows the proposed change after selection... (re @kc2bez: I am confused by that bug. Erase disk erases the disk. That wasn't the intention?) [16:11] I assigned it simply so we made sure to address it, but I tend to agree. [16:11] I mean, I'm all for UX making it easier to choose between the two. [16:11] That being said, aren't we just confusing the heck out of the average joe, even already with ext4? [16:11] That UX needs some cleanup, for laymen, while addressing this issue I thinl. [16:11] *think [16:11] Anyway, let's follow up on the bug too. :) [16:18] [telegram] I think the bug was more about the partition table so GPT vs MBR rather than the file system type but point taken about the UX there. [16:19] [telegram] I just need to get my head wrapped around it a little better so I can form an intelligent reply and appropriate questions. [16:19] Same page, would be the same idea I think, then went on to address the general point. :) [16:33] https://bugs.launchpad.net/ubuntu/+source/calamares/+bug/2046500/comments/3 [16:33] -ubottu:#lubuntu-devel- Launchpad bug 2046500 in calamares (Ubuntu) "Noble daily calamares installer does not honor existing partition table on 'Erase All'" [High, In Progress] [16:35] Destroying the partition table could be tantamount to suicide for the system if, e.g., an EFI system, GPT -> MBR. [16:36] [telegram] Agreed, it should never go in that direction. [16:37] Same in a system that doesn't do EFI going MBR -> GPT. It'd be useless. [16:39] [telegram] Right. Erasing the current table and recreating it seems like what "erase" should be doing. The type offered should be what is compatible with the mode the system is booted in. [16:44] Exactly. [18:23] re Cala bug: I came, I saw, I commented. [18:25] I also think the bug report is very odd. He mentions an MBR partition table being preserved on BIOS systems, which I both highly doubt, and which I think would cause boot issues. [18:26] er, an MBR table preserved on EFI systems I mean [18:28] [telegram] Marked as Won't Fix given all our discussion. Left a slightly salted comment saying as such heh [18:32] [telegram] arraybolt3: I was going to write a similar comment to yours but you already did it :) thanks for saving us time ;) [18:32] Glad to! [19:35] [telegram] Thanks for your comment arraybolt3 and thanks for closing the loop @tsimonq2 You both wrote it more eloquently than I could. [19:35] [telegram] Thanks to you as well @Eickmeyer [19:35] [telegram] Quite welcome. [20:25] tsimonq2: lmk whenever you've tagged the new release of installer-prompt, my time today is limited and I'd like to help get that pushed out today before ~4:40 PM [20:25] arraybolt3: It's all good now? [20:25] it's tagged? [20:26] oh, probably only on the git.l.m side [20:26] no I'm actually asking heh [20:26] forgot we had tags there for a second [20:26] tsimonq2: ah. Yes, it should be good now. [20:26] I verified that whatever was locking up wasn't (directly) my fault. [20:26] arraybolt3: Kay, doing the button pressing. Do you have upload access to that package in Ubuntu? [20:26] believe so [20:27] (ugh, over here ubuntu-upload-permission says no! [20:28] so nevermind, I can't upload there. [20:28] arraybolt3: One last question for you, this'll be quick. [20:28] Given semver, what version number shall this have? [20:29] 1.1.0? The behavior changed in major ways but not in ways that one might consider "breaking" IMO. [20:30] arraybolt3: I'm going to blindly take your word for it. Final answer? [20:31] * arraybolt3 checks the git log just to make very sure [20:32] unless you consider the inability to change locale after installation starts (and thereby kill the installation) a breaking change, yes, final answer. [20:32] otherwise there's no breaking new features. [20:32] Good enough for me. Thanks :) [20:36] https://github.com/lubuntu-team/installer-prompt/releases/tag/1.1.0 [20:36] 🎉 [20:51] Oh, right, I was gonna upload that XD [20:51] got sidetracked [20:53] Running it through sbuild real quick to make sure Lintian doesn't shriek. [20:54] Lintian: pass [20:55] Fully clean. Impressive work :) [20:55] woot [20:57] arraybolt3: By the way, how long *do* you actually drop off the face of the earth every week? :P [20:59] 24 hours. From sunset on Friday to sunset on Saturday CST. [20:59] Got it. No worries, I'll claim "coverage" during then. [20:59] Was just wondering if you'd be around for the ISO respin. [21:00] (I mean, if Britney is fast, you just might be. That being said, I respect your wishes and won't ask you to push it; in fact, that's honestly a great idea.) [21:00] oooooh openMSX just passed :D [21:00] going to check Lintian and then shoot that your way [21:02] sweeeeeeeeeeeeet [21:03] [matrix] gah, MoM mangled it [21:03] [matrix] glad I checked [21:03] [matrix] either that or else some of my changes weren't accepted by the maintainer... [21:03] "Always check on MoM." [21:03] [matrix] that's more likely actually [21:04] [matrix] (mainly the Lintian overrides were dropped for some false-positives) [21:04] [matrix] (probably since the maintainer wants the actual Lintian bug itself to be fixed) [21:04] [matrix] (which means I have to learn Perl *at some point*) [21:05] One Does Not Simply "Learn Perl" muahahaha [21:06] (sigh, the maintainer also didn't make a change I told him to) [21:06] that's OK though, I'll make it here. [21:14] tsimonq2: https://bugs.launchpad.net/ubuntu/+source/openmsx/+bug/2044565 [21:14] -ubottu:#lubuntu-devel- Launchpad bug 2044565 in openmsx (Ubuntu) "Merge openmsx from Debian" [Undecided, New] [21:17] [telegram] I likely will not get to this before you EOW, unfortunately :( [21:17] [telegram] I'll handle it tonight though