[02:10] <arraybolt3> juliank, mkukri: Looking at grub2/debian/grub-common.service. Is there any reason in particular this file embeds shell code and uses ExecStartPre, ExecStart, and ExecStartPost? Or is it acceptable to put all that into a script of some sort and have it called by ExecStart?
[02:11] <arraybolt3> Asking because I feel cramming the code to detect when grubenv is on the ESP into those lines is... decidedly non-ideal. I mean it's doable, but it will look like a disaster.
[04:13] <arraybolt3> At any rate, I have most of the code written (with some review help from mmikowski, thank you!) to enable putting grubenv on the ESP when running on an EFI system where grubenv wouldn't usually be writable. The last thing to fix is the grub-common.service. Anyway, going to go afk for probably the rest of the night, will catch up in the morning and see what needs done.
[07:53] <tjaalton> anyone else running noble and a thinkpad, and like to scroll with the middle-button pressed and using the nipple up/down, left/right? I'm seeing an infuriating issue where it send an extra middle-click when you stop scrolling, which means pasting whatever is on the buffer or opening new tabs in tbird or closing tabs on ffox..
[08:00] <mkukri> arraybolt3: did you take into consideration the case where multiple ESPs exist?
[08:01] <mkukri> for example when they are synced across two disks of a raid.  i believe that could cause problems because you dont know which one you booted from and which one contains the up to date grubenv
[08:32] <tjaalton> my scrolling issue seems to only happen with a thinkpad usb keyboard, not a real system, and it can be "fixed" for a while by unplugging the keyboard. but it comes back after some time.. and this started after upgrading to noble. dunno who to blame, kernel/libinput/gnome
[08:32] <tjaalton> "real system = thinkpad laptop"
[10:23] <tjaalton> or maybe it's just going bad, so sad.. an older model is working fine
[12:01] <bdrung> @pitot in
[12:01] <bdrung> @pilot in
[14:58] <xypron> \o
[15:00] <ginggs> xypron: ECHAN?
[15:05] <ogra_> ginggs, yoga fits every chan 😛
[15:35] <arraybolt3> mkukri: I did not take that into account, thanks for mentioning that! I figured this was going to be edge-case prone and was actually thinking of other edgecases (complex multiboot, what if the user wants to write grub.cfg manually, potential for grubenv file format to change, what if GRUB is in EFI/BOOT rather than EFI/ubuntu, etc.). Hopefully by the time I have a debdiff out it'll be close
[15:35] <arraybolt3> to usable.
[15:35] <mkukri> arraybolt3 please file this as a git MP instead of debdiff
[15:35] <mkukri> https://git.launchpad.net/~ubuntu-uefi-team/grub/+git/ubuntu
[15:36] <juliank> One note
[15:36] <juliank> We have a primary ESP, mounted to /boot/efi. We can record the uuid of that
[15:36] <juliank> then in the grub.cfg you can search --fs-uuid=<uuid of esp>
[15:36] <juliank> to then find the grubenv in that
[15:37] <juliank> If your primary ESP fails, you lose the environment but merh
[15:37] <arraybolt3> yep, that's what I'm doing
[15:37] <arraybolt3> juliank: oh, awesome
[15:37] <arraybolt3> mkukri: thanks for the git link, I wondered if one of those existed
[15:37] <arraybolt3> also, about grub-common.service - any reason to not yank that logic into its own script and just call that?
[15:37] <mkukri> there is also a repo under ~ubuntu-core-dev which is technically the official one, but for reviews that one is fine
[15:37] <arraybolt3> that will let me add the detection logic for when grubenv is on the ESP
[15:38] <arraybolt3> (There's already a script under /etc/init.d/grub-common that looks to do the same job, but that's apparently not being called by systemd since when I modified it, nothing happened in my test environment)
[15:38] <mkukri> i guess you can put a script in /usr/lib/grub/grub-your-script and call that
[15:38] <arraybolt3> kk
[15:39] <juliank> I think we removed /etc/init.d/grub-common
[15:40] <juliank> I thought we had removed it, anyway
[15:41] <mkukri> it exists on my noble vm, not 100% what package puts it there...
[15:41] <mkukri> i guess grub-common, but i think only the sd service works
[16:04] <sudip> can anyone please mark LP: #1970521 as affecting Jammy
[16:04] -ubottu:#ubuntu-devel- Launchpad bug 1970521 in boinc (Ubuntu) "boinc-client crashes when started with core dump on Xubuntu 22.04" [Undecided, Confirmed] https://launchpad.net/bugs/1970521
[16:05]  * sudip requested in #ubuntu-bugs yesterday but..  :(
[16:06] <arraybolt3> mkukri, juliank: running this idea by you - in the event of a multiboot scenario with multiple Ubuntu versions sharing the same ESP, every time one of the installed Ubuntu versions "takes over" the ESP, it will change the GRUB version there but *not* the grubenv file. That could potentially result in grubenv being too old or too new for the installed GRUB. The easiest way I can think of to
[16:06] <arraybolt3> get around this is to simply nuke and recreate grubenv on every update-grub event (which I assume will happen on every GRUB installation event, though I guess it might not, so perhaps the nuke-and-recreate has to be done in GRUB's postinstall too?).
[16:06] <arraybolt3> sudip: happens on Noble and Jammy?
[16:06] <sudip> nope, just Jammy please
[16:07] <arraybolt3> sudip: .
[16:07] <sudip> thanks arraybolt3
[16:07] <arraybolt3> glad to help :)
[16:10] <bdrung> @pilot out
[16:59] <kanashiro> @pilot in
[17:21] <arraybolt3> mkukri: well I think my patch is finished, *but* I can't get GRUB to build because of a Python unit test failure in test_grub_sort_version.py where grub-sort-version errors out with "ModuleNotFoundError: No module named 'apt_pkg'".
[17:21] <arraybolt3> smells like something transition-related to me, not exactly sure.
[17:24] <mkukri> arraybolt3 ah crap it's not, that repo has unpublished changes, and that's a bug not caught on my box, please add python3-apt to the build-depends in d/control and it should built (it's already in regular depends but in needs to be in both with those tests)
[17:25] <arraybolt3> aha, that makes sense (I didn't spot it because it was in Depends).
[17:27]  * mkukri has some hacky "speed up grub build" chroots that has the depends and build-depends preinstalled
[17:28] <arraybolt3> oooh nice, I'm over here just using plain-ol sbuild
[17:28] <arraybolt3> though I do make use of shm overlays
[17:28] <mkukri> it's a regular sbuild chroot that i just schroot into and preinstall stuff for re-running sbuild
[17:29] <arraybolt3> oooh nice
[18:06] <ahasenack> sudip: I'm really enjoying your SRU verifications
[18:06] <ahasenack> in a sea of ambiguous verifications that we see many times, yours are very precise
[18:42] <arraybolt3> It looks like my patch is working!
[19:21] <sudip> ahasenack: :)
[19:22]  * sudip still has some pending, which is in my weekend's todo list
[19:23] <sudip> and sagemath is still pending.. :(
[19:23]  * sudip wonders if waveform will want to enjoy another round of sagemath
[19:33] <arraybolt3> mkukri, juliank: https://code.launchpad.net/~arraybolt3/grub/+git/grub/+merge/462445
[20:10] <adrien> does anyone know how to edit a PPA's dependencies to include -proposed? I can see --ppa-dependencies but that seem to be only for dependencies to other PPAs
[20:10] <adrien> bryceh: ^
[20:13] <arraybolt3> adrien: https://ibb.co/S6k2NvD
[20:13] <adrien> arraybolt3: I'd like to do that from ppa-dev-tools instead
[20:14] <arraybolt3> ah, sorry
[20:14] <adrien> looking at the code, it doesn't seem possible but I'm not familiar with the code
[20:19] <kanashiro> @pilot out