[00:14] [matrix] Maybe it is just me but I had an issue installing a web browser on a minimal install. I didn't try on a normal install yet. I tried Falkon, qutebrowser, and Gnome web since I knew they all were in the archive. They all seg fault at start. [00:15] [telegram] Oh *wow*. Okay, nice. Something's not pulling in the right backends... [00:16] [telegram] It seems like, I haven't narrowed it down yet. I thought I would mention it so we could all investigate. [00:43] [telegram] Can confirm this on the Kubuntu daily ISO [00:49] [telegram] Oh fun, that stretches beyond where I was looking. [00:49] [matrix] sounds like a library changed wrong [00:56] [matrix] does the build tooling ever "forget" to do a soname bump or something? [01:00] [telegram] qutebrowser may be easiest to look at, as it has the fewest dependencies [01:04] [telegram] Could potentially just be an NCR [01:05] [telegram] worth trying in a ppa [01:05] [telegram] I remember qutebrowser working somewhat recently..... [01:07] [telegram] the whole FORTIFY_SOURCE=3 thing is also making some things go weird [01:08] [telegram] though I don't immediately see something common between these recently rebuilt with that [01:21] [telegram] 💤🛏 time [02:12] I wonder if the browser nonsense potentially has the same root cause as the riscv64 failure mess [06:44] teward: rearranging power levels? [06:44] * arraybolt3 just got the email that I now have ISO rebuild rights [14:32] [telegram] for Qt based apps, seems like this crash could be affecting anything that uses qtwebengine [14:33] [telegram] ghostwriter, tellico, notepadqq all crash on start [14:33] [telegram] and those are only the 1st 3 I tried [14:36] [telegram] qtwebengine hasn't been rebuilt in noble for nearly a month, so something else has changed to make it misbehave [14:39] [telegram] /me looks suspiciously at ICU [14:57] [telegram] https://matterbridge.lubuntu.me/b221e1c5/file_10265.jpg [14:57] [telegram] Forwarded from RikMills: ^^ ghostwiter [14:57] [telegram] https://matterbridge.lubuntu.me/00dce717/file_10266.jpg [14:57] [telegram] Forwarded from RikMills: falkon ^^ [16:16] [telegram] where is calamares packaging now? [21:25] did picom-conf already get seeded? [21:25] actually, /me checks [21:26] it is not, I'll add it [21:27] [telegram] Ack jfdi [21:28] [telegram] +1 [21:28] and added [21:28] now to regen the lubuntu-desktop metapackage [21:29] uh oh, meta was last updated in... Lunar?! [21:29] * arraybolt3 fixes [21:30] ah, no, just hasn't been uploaded to Git. [21:36] tsimonq2: just to be sure I remember how to do this, I just run the "update" script, fix the changelog, and then that's it, right? [21:36] (and then of course upload to Git and LP) [21:38] [telegram] That sounds correct [21:38] for some reason when I do that, Git asks me for my username and password to LP even though I have SSH configured I thought... [21:39] I mean I literally just cloned and pushed over SSH a second ago updating the seeds themselves. [21:39] [telegram] Yeah, LP is a little different [21:39] I was pulling and pushing to/from LP [21:39] with SSH, successfully [21:39] so idk what it's doing [21:40] I know for a fact I used to have SSH pull/push working here. [21:41] [telegram] Gotcha [21:41] I think it's an ssh config issue... lemme try adding the server explicitly. [21:41] oh, of course it is [21:41] because in my VM, my username is "user", not "arraybolt3" [21:41] [telegram] Makes sense [21:41] so it's trying to ssh in as "user" with the wrong key [21:46] got it working [22:07] [matrix] hmm, picom-conf isn't in the archives yet it appears === guiverc2 is now known as guiverc [23:01] arraybolt3: Yeah, you have it right :) [23:01] Alright guys, *where* is the critical bug? [23:01] "But Simon, we haven't found anything recently?" [23:02] No, I can feel it. There's one somewhere. It has been raining *too* hard for it not to be pouring. [23:02] Show yourself, bug. :P [23:04] ICU or qtwebengine? [23:05] The latter takes like a day to build on a fast machine, I'm not excited for that :P [23:05] Yeah, I don't think anyone is [23:05] You asked for a critical bug, not an easy one :P [23:06] I'm willing to build it on my machine though if it would be helpful. [23:06] Alternatively we could just NCR it and see what happens [23:06] Rik left some screenshots above [23:09] Both qutebrowser and ghostwriter launch without issue on my system... [23:10] 🤔 [23:10] weird [23:10] falkon? [23:11] tsimonq2: https://bugs.launchpad.net/ubuntu/+source/falkon/+bug/2046796 [23:11] You've updated, right? [23:17] Doesn't launch in a brand new Noble VM over here. [23:18] Falkon: Creating new profile directory [23:18] Error creating database schema "No query Unable to fetch row" [23:18] [1712:1712:1218/181615.678815:FATAL:credentials.cc(125)] Check failed: . : Permission denied (13) [23:18] Trace/breakpoint trap (core dumped) [23:18] Fully updated noble system here. [23:18] Might be graphics hardware related? [23:19] $ apt list --installed | pastebinit [23:19] https://dpaste.com/CYHXNFRXM [23:20] Same version of qtwebengine over here [23:21] maybe I have extra qtwebengine packages that the default install does not? [23:26] https://dpaste.com/ESFQWEMD8 [23:26] looks like we have the same webengine packages [23:27] I wonder if this is a SIGILL? [23:29] hrm, kernel message is pretty cryptic [23:30] kernel: [ 2760.894641] traps: falkon[1712] trap int3 ip:7f50be9d7b13 sp:7ffc7a5755a0 error:0 in libQtWebEngineCore.so.5.15.15[7f50bc7b4000+6931000] [23:30] I don't know what a breakpoint trap is... [23:30] curious that epiphany creashes with the same Trace/breakpoint trap (core dumped) [23:31] despite not using qtwebengine [23:31] It's almost like something got built in debug mode? [23:31] And had a breakpoint left...? Is that even possible? [23:38] Epiphany sigtrap is in g_log_structured_array in libglib-2.0 [23:39] https://i.imgur.com/15jwUUz.png [23:40] oh interesting, look at this: [23:41] https://i.imgur.com/3XUUKMv.png [23:41] I managed to get a SIGILL out of Falkon after the SIGTRAP [23:41] [telegram] https://matterbridge.lubuntu.me/1457f9af/file_10269.jpg [23:42] wait so there's literally a breakpoint defined in the code? [23:42] something to do with logging in qt and gtk apps [23:44] trying falkon with debuginfod now [23:44] this is certainly the weirdest error of this sort I've ever seen [23:46] the last week or 2 seems to have been full of wierd crashes and bugs [23:52] debuginfod seems to have done almost nothing helpful (didn't get any fancy code snippet like you did), but both the SIGTRAP and the SIGILL are happening in that LogMessage() destructor. [23:52] RikMills: Are you working in a VM or on real hardware? [23:53] If in a VM, are you using host CPU passthrough? If not, could you try that? [23:53] I have a distant thought in the back of my mind that maybe somehow a setting set for the x86_64-v3 rebuild of Lunar somehow propagated to Noble on accident. [23:54] not sure if I can set that in virtualbox [23:54] hrm, I'll try it on my end then