| 3 Apr 2022 |
 flokli | Seems https://github.com/NixOS/nixpkgs/issues/167001 did regress containers | 06:54:51 |
| flokli | * Seems https://github.com/NixOS/nixpkgs/pull/165453 did regress containers: https://github.com/NixOS/nixpkgs/issues/167001 | 06:55:34 |
| flokli | I'm in the bike today, but Janne Heß can you take a look? | 06:56:27 |
 bobvanderlinden | In reply to @andreas.schraegle:helsinki-systems.de
If you want to look into it, this commit dropped the generator and updated a bunch of documentation. Aaah we probably just have to use the lvm2_dmeventd package. | 07:06:23 |
 @janne.hess:helsinki-systems.de | In reply to @flokli:matrix.org
I'm in the bike today, but Janne Heß can you take a look? Ah I just saw the message... It's fixed | 13:42:36 |
 @elvishjerricco:matrix.org | Words can't describe how happy I am that this has been merged :) I've been messing with this for over two years, trying to find the right way for it to be merge-worthy. And now it's there and it's awesome. | 17:40:06 |
 Arian | Yay! | 17:59:57 |
| Arian | I'll be spending ocean sprint on a lot of the follow up tickets I think | 18:00:04 |
| Arian | Luks, Plymouth, silent boot | 18:00:15 |
| Arian | But I forgot to bring my non-mac laptop so maybe I'll work on M1 support on Nixos instead lol | 18:00:47 |
| flokli | Redacted or Malformed Event | 18:26:07 |
| flokli | Redacted or Malformed Event | 18:26:19 |
| *  colemickens 👀 for a luks pr to test | 23:18:11 |
| 4 Apr 2022 |
| @janne.hess:helsinki-systems.de | ElvishJerricco or bobvanderlinden: didn't one of you try to build the image without calling cpio? Is that still an ongoing effort? | 09:25:44 |
| @elvishjerricco:matrix.org | Janne Heß: How would you make the archive without cpio? Maybe you're thinking of what bobvanderlinden did to avoid copying all the files to a temporary location before writing out with cpio? | 09:27:21 |
| @elvishjerricco:matrix.org | That's something I'd consider valuable but never bothered to add to the PR | 09:27:40 |
| @janne.hess:helsinki-systems.de | I thought you were planning to use some tool that fell out of the kernel source tree… 🤔 | 09:27:48 |
| @elvishjerricco:matrix.org | I do not know of such a tool | 09:28:12 |
| @janne.hess:helsinki-systems.de | maybe I am just mistaken between the hundreds of messages here :D | 09:28:23 |
| @elvishjerricco:matrix.org | Janne Heß: Maybe it was this? https://github.com/NixOS/nixpkgs/pull/120015#issuecomment-1066027673 | 09:55:32 |
| @janne.hess:helsinki-systems.de | ah probably, yes | 09:57:42 |
| @janne.hess:helsinki-systems.de | but we worked around the issue by using the rust tool? | 09:57:51 |
| @elvishjerricco:matrix.org | Janne Heß: Don't think so. The rust tool still copies stuff around. | 10:21:16 |
| @elvishjerricco:matrix.org | bobvanderlindenmade a commit a while ago that avoided the copying | 10:22:40 |
| bobvanderlinden | Indeed. It was in a fork of your old branch. Outputs paths on stdout and pipes that to cpio.
It also included looking for other /nix/store paths in non-elf files, but that part was considered unneeded and may be dropped.
I can take a look creating a PR for the change tonight (20.00 cest). | 10:35:05 |
| @elvishjerricco:matrix.org | bobvanderlinden: Yea I'd very much like a PR for the part that avoids copying, but the part that searches for other /nix/store paths was probably a problem for a different PR | 10:37:44 |
| bobvanderlinden | Agreed 👍 different discussion | 10:38:23 |
| bobvanderlinden | It did still use cpio though. Might be cool to implement a basic cpio output in the rust tool as well, but that's also something for later I guess | 10:39:35 |
| @elvishjerricco:matrix.org | I have no problem with using cpio. Avoiding the copying is valuable because it produces less wear and tear on the underlying storage | 10:41:05 |
| @elvishjerricco:matrix.org | I know it's minimal, but for stuff like Hydra builders that will make a bazillion of these per day, it can make a difference | 10:41:44 |
