| 2 Jun 2023 |
 trofi | Normally nix should work the same on gentoo. The impurity that usually matters is a host kernel, filesystem type and nix version itself. | 06:34:00 |
 artemis | so per that,
host kernel is 6.1.27
filesystem is zfs
nix version is 2.15.0 | 06:38:43 |
| artemis | but | 06:39:43 |
| artemis | i was gettin the problem on ext4 too so its probably not that. (we did a migration from ext4 to zfs in between working on this) | 06:40:05 |
| artemis | * i was gettin the problem on ext4 too so its probably not zfs. (we did a migration from ext4 to zfs in between working on this) | 06:40:23 |
| artemis | kernel is gentoo's sys-kernel/gentoo-kernel config which is a pretty bog standard kitchen sink kernel like u get on most distros | 06:42:13 |
| trofi | Any unusual mount options? | 06:43:36 |
| artemis | asdf i just realized we literally use yr overlay to install nix thats vry funny. thought u seemed familiar | 06:44:11 |
| artemis | no, nothing funny with the mount | 06:44:21 |
| artemis | philomena/rootfs on / type zfs (rw,relatime,xattr,posixacl) | 06:44:53 |
| trofi | yeah, looks fine | 06:45:05 |
| trofi | :) | 06:45:17 |
| artemis | [tangentially, aarch64 build was broken for a bit due to boost but they bumped boost and it works again now] | 06:45:47 |
| artemis | i also imagine that its not just you for whom this is working just fine considering i initially ran into this using the exact flake.lock in snektron's repo and i have to assume the build worked fine for them. definitely really weird | 06:49:05 |
| artemis | that and i dont get the problem when building libtool for x86 | 06:49:52 |
| artemis | * im gonna see if i run into this cross compiling to something that isnt riscv | 06:51:39 |
| artemis | yup, get it whel building for pkgsCross.aarch64-multiplatform | 06:55:08 |
| artemis | * yup, get it when building for pkgsCross.aarch64-multiplatform too | 06:55:13 |
| trofi | Yeah. It's a bit worrying that gentoo stopped using gcc releases and now uses snapshots frm stable branches. That literally makes their gcc to emit different code from everyone else. For things like kernel it's such an opportunity to get uniquely broken kernel. I hope it's not it, but making working/broken comparison is hard against gentoo. | 06:55:48 |
| trofi | Reproduced libtool breakage in a gentoo chroot \o/ (on a nixos kernel). | 06:56:40 |
| artemis | neat!. also i didnt know that about gcc | 06:57:01 |
| trofi | https://packages.gentoo.org/packages/sys-devel/gcc - all green boxes (stable) are date-suffixed. | 06:57:32 |
| artemis | oh yeah. why do they do that i wonder | 06:57:57 |
| trofi | to avoid backporting known fixes (and bugs) that were already backported :) | 06:58:36 |
| trofi | If you add sandbox-fallback = false to /etc/nix/nix.conf, do nix build commands start failing for you? | 07:40:45 |
| artemis | added that, ran rc-service nix-daemon restart, there doesn't seem to be a change in behavior in `nix build` | 07:54:07 |
| artemis | rebuild of x86 host libtool worked still. build of cross libtool failed still | 07:54:37 |
| trofi | nod. I'll try to figure out why sandbox gets disabled for me and get back to you later. | 08:01:59 |
| artemis | ^^ thanks for taking a crack at it | 08:05:41 |
 misuzu | In reply to @artemis:arty.chat
though. DHCP never succeeds for some reason so i am forever stuck with the DHCP 169.whatever address You probably have v1.3B board but image is configured for v1.2A. You could try mine prebuilt image for v1.3B boards: https://github.com/misuzu/nixos-vf2/releases | 08:09:39 |
