30 Sep 2023 |
lovesegfault | Artturin: when you're around, lmk what needs uploading, I'll make time to get everything uploaded | 17:54:55 |
Artturin | In reply to @lovesegfault:beme.ems.host Artturin: when you're around, lmk what needs uploading, I'll make time to get everything uploaded All the linux bootstrap files have to be updated
mips64el-n32.nix
loongson2f.nix
mips64el.nix
armv5tel.nix
armv6l-musl.nix
armv6l.nix
armv7l.nix
i686.nix
mipsel.nix
powerpc64le.nix
riscv64.nix
x86_64-musl.nix
x86_64.nix
aarch64-musl.nix
aarch64.nix
| 18:08:57 |
Artturin | which is why i'd like to make a script to help you | 18:09:14 |
Artturin | https://github.com/NixOS/nixpkgs/issues/253713#issuecomment-1712373087 | 18:09:19 |
Artturin | * https://github.com/NixOS/nixpkgs/issues/253713#issuecomment-1712373087 please check | 18:09:37 |
lovesegfault | Heh, yeah, we'll need some automation for that to be viable, for sure | 18:09:41 |
Artturin |
in #151399 (comment) you asked for the sha256sum of all the on-server components, however shouldn't it be okay to directly upload the packages from bootstrapFiles
| 18:10:04 |
lovesegfault | Let me grab my notes on the process, one second | 18:10:05 |
Artturin | i can work with the notes to make a script | 18:10:19 |
lovesegfault | In reply to @artturin:matrix.org
in #151399 (comment) you asked for the sha256sum of all the on-server components, however shouldn't it be okay to directly upload the packages from bootstrapFiles
Yup, it should be, I ask for the sha as an extra verification step to be uber-sure I have the right thing | 18:10:43 |
lovesegfault | Because I can upload tarballs, but I cannot delete them | 18:10:51 |
lovesegfault | So I err on the side of being too cautious | 18:11:06 |
lovesegfault | Redacted or Malformed Event | 18:27:16 |
lovesegfault | Uploading tarballs
- Look at the Hydra build linked on GH, make sure nothing is off (built the
right thing, on the right pkgs intance/channel/etc)
- Check the build logs on Hydra, make sure there are no glaring issues that did
not end up failing the build
- Get the nix-store path, and sha256's from GH
nix build -L $storePath
- Get the
nixpkgs commit which generated the tarball
- Check it out, build it, make sure it's reproducible
- Figure out the correct subpath for this tarball
- x86_64-darwin -> stdenv-darwin/x86_64
- Check that the hashes within match the ones from GH
- Construct the upload path
- s3://nixpkgs-tarballs/$os/$arch/$nixpkgs_commit
- e.g. s3://nixpkgs-tarballs/stdenv-darwin/x86_64/05ef940b94fe76e7ac06ea45a625adc8e4be96f9
- Upload it
- e.g. aws s3 cp --recursive --acl public-read /nix/store/9h4d7s313wv3gkfwi493yr1wvdsz9lf2-stdenv-bootstrap-tools/on-server/ s3://nixpkgs-tarballs/stdenv-darwin/x86_64/05ef940b94fe76e7ac06ea45a625adc8e4be96f9
| 18:27:33 |
lovesegfault | Artturin: there ya go, that's what I do | 18:27:42 |
Artturin |
Look at the Hydra build linked on GH, make sure nothing is off (built the right thing, on the right pkgs intance/channel/etc)
Maybe we should merge https://github.com/NixOS/nixpkgs/pull/256439 after all, then we can get hydra builds for the bootstrap files
| 18:49:40 |
lovesegfault | In reply to @artturin:matrix.org
Look at the Hydra build linked on GH, make sure nothing is off (built the right thing, on the right pkgs intance/channel/etc)
Maybe we should merge https://github.com/NixOS/nixpkgs/pull/256439 after all, then we can get hydra builds for the bootstrap files
Sounds good, merged :) | 19:17:54 |
lovesegfault | I'm sorry for reverting that, I was having a chaotic week and didn't fully understand what the PR was doing. Now I see that it was AOK to have been merged and I was wrong | 19:18:37 |
lovesegfault | Can someone review this before I upload it? https://github.com/NixOS/nixpkgs/pull/255451 | 19:46:00 |
lovesegfault | trofi: Regarding https://github.com/NixOS/nixpkgs/pull/188544
Is there a flag I can pass for it to find the plugins?
| 21:39:31 |
trofi | --plugin should do it, but i don;t know what it does to existing plugins and how it handles multiple ones | 21:40:54 |
trofi | And the typical problem if trying to guarantee that both gcc and binutils use the same libstdc++.so library. | 21:42:00 |
lovesegfault | Interesting:
cc1: fatal error: plugin /nix/store/h5kvfrjmpw792v8jg7nrzfkffmn0iyy8-gcc-12.3.0/libexec/gcc/x86_64-unknown-linux-gnu/12.3.0/liblto_plugin.so is not licensed under a GPL-compatible license /nix/store/h5kvfrjmpw792v8jg7nrzfkffmn0iyy8-gcc-12.3.0/libexec/gcc/x86_64-unknown-linux-gnu/12.3.0/liblto_plugin.so: undefined symbol: plugin_is_GPL_compatible
| 21:45:09 |
lovesegfault | Ohhhhh | 21:49:35 |
lovesegfault | We must accidentally strip that symbol lol | 21:49:43 |
lovesegfault | hrm, I wish I could easily set a list of paths to exclude from stripping in the fixupPhase | 21:57:45 |
Artturin | In reply to @lovesegfault:beme.ems.host hrm, I wish I could easily set a list of paths to exclude from stripping in the fixupPhase You can | 22:01:48 |
Artturin | stripExclude | 22:02:00 |
Artturin | There was a pr which I rebased and merged a few weeks ago | 22:02:29 |
Artturin | Should be in master by now | 22:02:34 |