[18:25] [matrix] It's been a long while since we discovered that bug with lubuntu-update-notifier and config-file-prompt. I have a working fix, but so far we've not even started to roll it out and only barely reviewed it. [18:25] [matrix] I know Simon Quigley said we should probably be talking to the Technical Board about this, but I don't know where to start there. [18:26] [matrix] What are our next steps? Should I be writing an email? Pinging relevant people in #ubuntu-devel? Uploading what I have to Lunar and preparing an SRU, and we'll get the TB's input at that point? [18:26] [matrix] I think we should probably be getting the ball rolling on this sooner rather than later, as February looms closer, since we didn't want to delay the SDDM patch Eickmeyer made for too long. [18:28] [matrix] The fix is here: https://git.lubuntu.me/Lubuntu/lubuntu-update-notifier/src/branch/arraybolt3/upgrade_repair The problem was that config-file-prompt in the repair function doesn't work, and from the looks of the aptd code it won't work without changes to aptd itself. [18:30] [matrix] I was going to ask about this the other day. The SDDM update can't remain on hold indefinitely. Deadline is Thursday and I'm unblocking it. I can't wait any longer. [18:30] [matrix] The only reason we've not been moving faster is that Simon Quigley has been unavailable for a while and we (or at least I) was waiting on his input. [18:31] [matrix] Yeah, but I can't wait on him forever. [18:31] [matrix] True. [18:31] [matrix] Thus why I'm asking what we should do now, hoping that someone who knows more about what he's asking for will be able to help me get things going forward again. [18:32] [matrix] This is also why I'm imposing a deadline. Gotta light a fire somewhere. [18:33] [matrix] Eickmeyer: Do you know what the best way for contacting the Technical Board is? [18:33] [matrix] technical-board@lists.ubuntu.com [18:33] [matrix] In this situation, I'm thinking, I should upload a fix to Lunar, prepare the SRU, specify that I would like a member of the Technical Board to review it and give input on the situation, and then move forward. Lunar can always be repaired. [18:34] [matrix] Honestly, I don't know why the TB needs to be involved. [18:34] [matrix] Maybe I should just do both - prepare the SRU and send an email. [18:35] [matrix] I thought we already had a fix in lunar? [18:35] [matrix] Dan Simmons : Not for the case in which an upgrade gets interrupted and breaks. [18:36] [matrix] Which is the heart of the problem in stable releases - no matter what we do, SDDM going through *will break* all stable Lubuntu releases. We can't avoid that breakage, so we need a patch to repair it after it happens. [18:36] [matrix] That's what my proposed fix is, but it's not quite perfect and can't be made perfect due to an aptd limitation - when you use fix_incomplete_install(), the config_file_conflict event silently fails since the command used by aptd doesn't generate config file conflict prompts. [18:38] [matrix] If SDDM goes through and we don't have the patch out, all of our non-command-line users will have their systems unable to upgrade. At least if our fix is out before SDDM goes through, because it only reqiures that lubuntu-update-notifier get unpacked, not configured, and unpacking always happens before configuration, it will work. [18:39] [matrix] So if the fix can be pushed out even a tiny bit before SDDM goes through, we're good. But now that we don't have time for even a full week-long SRU verification point, we may still end up with a *lot* of users stuck in a lurch. [18:39] [matrix] So if the fix can be pushed out even a tiny bit before SDDM goes through, we're good. But now that we don't have time for even a full week-long SRU verification period, we may still end up with a _lot_ of users stuck in a lurch. [18:40] [matrix] (I'm hoping to be able to get things fixed today and then beg Eickmeyer to give us more time than the deadline he set, but I don't want to ask that now since, if I can't get it done today, then what?) [18:40] [matrix] +++++++++++++++++ [18:40] [matrix] Sorry, dog got my keyboard [18:40] [matrix] lol I was just about to ask if that happened :P [18:41] [matrix] Look, I'll go for next week. But this has gone on for way too long. [18:41] [matrix] Dan Simmons : I don't want to just "take control" and fix things without approval from other developers. Is it OK if I get the current workaround uploaded and begin the SRU process? [18:42] [matrix] I think we need to go, yes. [18:43] [matrix] K. I'll finish fixing up the thing to get rid of config-file-conflict since it doesn't work anyway, do some final testing, and then try to get it uploaded and out there. [18:44] [matrix] Might get some coffee and food while I'm at it :D [18:45] [matrix] 👍️ [19:57] Got the latest iteration uploaded to a PPA, if this works then the deployment process can begin. [19:57] (And also I pushed it to Git.) [20:28] [matrix] Bug fix works on Jammy. [20:28] [matrix] Next step is probably to test on Kinetic, I don't think Lunar will need testing. [20:29] [matrix] And I'm not sure how to test with Focal... I guess I could make a dummy package that trigges config-file-conflict? [20:29] [matrix] And I'm not sure how to test with Focal... I guess I could make a dummy package that triggers config-file-conflict? [21:51] [matrix] Eickmeyer: Wasn't that SDDM update still in -proposed in Kinetic? [21:51] [matrix] For some reason my Kinetic test VM just updated without problems rather than crashing and burning on the first upgrade attempt. [21:52] [matrix] (Which sounds good but it probably means the update either isn't in Kinetic, or it migrated and I installed it from the terminal, in which case 🫠) [21:52] [matrix] Yes, but only because I threw a ```block-proposed``` tag on it. [21:53] [matrix] Basically, I'm keeping it from migrating. [21:54] [matrix] Hmm, you're right, I guess it just... doesn't conflict with anything in Lubuntu upon upgrade? [21:54] [matrix] Meh, I guess that's actually a good thing, so I'll test this some other way then. [21:55] [matrix] Hmm, you're right, I guess it just... doesn't conflict with anything in Lubuntu **Kinetic** upon upgrade? [21:55] [matrix] (It crashes as expected on Jammy and the fix lets it recover, to be clear.) [21:57] [matrix] Probably because something else was released that causes the same issue. [21:58] [matrix] It only goes wonky when -proposed is enabled, and it's SDDM it always was hanging up on. [21:58] [matrix] Maybe the config file in question just didn't exist on Kinetic. [21:59] [matrix] Anyway, it's of no consequence, the bugfix is still going well, if it passes this test it will be upload time. [22:25] [matrix] Alright it works! \o/ [22:25] [matrix] Currently writing up the second bug report / SRU template. [23:04] [matrix] Sheesh, that took forever. I present for your consideration, the following horror story of a bug report: https://bugs.launchpad.net/ubuntu/+source/lubuntu-update-notifier/+bug/2002255 [23:04] -ubot93:#lubuntu-devel- Launchpad bug 2002255 in lubuntu-update-notifier (Ubuntu Lunar) "lubuntu-update-notifier provides no good way to resolve a forcibly terminated system update" [High, In Progress] [23:04] [matrix] Eickmeyer, Dan Simmons : ^ [23:05] [matrix] Let me know what needs changed, if anything. [23:05] * lubot [matrix] cracks knuckles and uploads the fix to Lunar [23:11] [matrix] And that's it for Lunar! \o/ [23:12] [matrix] Now it's just the SRUs to handle. [23:38] [matrix] arraybolt3: LGTM [23:50] [matrix] Gonna wait for everything to finish clearing through Lunar's -proposed queue, then upload to Kinetic and Jammy. [23:50] [matrix] Oh, I should merge this into ubuntu/lunar in Lubuntu Git. [23:58] [matrix] Got everything primed for the SRU now Git-wise, as soon as it's time to upload it's just a matter of a couple of debuilds and dputs. [23:58] [matrix] (And git checkouts and ...)