| 24 Mar 2025 |
 @elvishjerricco:matrix.org | That should be a separate directory | 19:01:53 |
 p14 | Ah, that'd probably be it. I was operating under the belief that this would be the upperdir from before this one was mounted, not so, eh? | 19:05:35 |
| p14 | Looks like did the trick ElvishJerricco , thank you for rescuing me from my own insanity <3 | 19:10:50 |
| @elvishjerricco:matrix.org | it would be if you were just running mount manually. But the overlay module in nixos creates systemd dependencies on the file systems for the upper/work/lower dirs, so I think you were creating a cyclic dependency of /nix/store on itself which probably blew something up | 19:12:18 |
| @elvishjerricco:matrix.org | i.e. I think it probably resulted in an fstab line like overlay /nix/store overlay ...,x-systemd.requires-mounts-for=/nix/store | 19:13:12 |
| @elvishjerricco:matrix.org | I guess it would work if you didn't use the nixos overlay module and instead just manually wrote the options | 19:13:54 |
| p14 | Well it didn't work great when I did that either. At least having the writable stuff go to a different place is working well enough | 19:14:48 |
| p14 | Now for my next challenge, make my x86 image also legacy bootable 🤦. I'd love to use the systemd-repart image builder for this but maybe that is too much to ask. | 19:15:52 |
| p14 | I'm wondering if I can get away with building an image with the grub MBR and use repart to put that image in the right place. But then I'm not really clear on whether I can mark this partition as bootable, I'm guessing not since this is not the systemd way. | 19:16:50 |
| p14 | (the systemd way being UEFI boot, because, obviously) | 19:17:00 |
| @elvishjerricco:matrix.org | p14: so, repart will not mark a partition as LegacyBIOSBootable, in fdisk speak. That's a GPT flag though, and I don't think it's strictly necessary for BIOS booting; I think the MBR just has to be done right. Nonetheless, I do have a repart patch for allowing setting that GPT flag with its numeric identifier: https://gist.github.com/ElvishJerricco/6059640087e1ee09744bb9df8f18bb1a | 19:22:34 |
| @elvishjerricco:matrix.org | but, getting an MBR configured properly for the image using repart is probably nontrivial and I have no idea how to do that | 19:23:08 |
| @elvishjerricco:matrix.org | well, I mean I know repart won't help | 19:23:22 |
| @elvishjerricco:matrix.org | but I mean letting repart make the GPT image, and then separately writting the MBR on LBA 0 | 19:23:57 |
| @elvishjerricco:matrix.org | no idea how to do that for grub | 19:24:01 |
| p14 | Is it not sufficient to supply a --image pointing to a pre-made grub image? I'm not aware of all the wiring details. | 19:29:52 |
 Arian | I have code for making a hybrid image with repart and runInLinuxVm lying around | 19:43:50 |
| Arian | I stole the code from mkosi | 19:44:24 |
| @elvishjerricco:matrix.org | I'd really like to be able to install grub on an image without a VM, but oh well | 19:44:46 |
| Arian | Yeh this I was not able to get working unfortunately | 19:45:08 |
| 11 Apr 2025 |
|  mjm changed their profile picture. | 22:23:09 |
| 24 Mar 2025 |
| p14 | I was hoping to at least make that part a small image that could be cached. WDYT, Possible? | 20:45:27 |
| Arian | Ah and then copy the bytes over? | 20:58:50 |
| 11 Apr 2025 |
|  @oxalica:matrix.org left the room. | 23:31:19 |
| 24 Mar 2025 |
| Arian | Yeh could be doable | 20:58:53 |
 emily | this is not a desire anyone should have :p | 21:37:46 |
| 12 Apr 2025 |
|  @NYXT:matrix.org left the room. | 02:00:31 |
| 24 Mar 2025 |
| @elvishjerricco:matrix.org | I mean I'd also like to be able to do that for our systemd-boot stuff | 21:39:13 |
| 12 Apr 2025 |
|  rappet changed their display name from Lucy to rappet. | 12:12:01 |
| 24 Mar 2025 |
| emily | that one seems much easier to acheive? | 21:40:46 |
