| 30 Sep 2023 |
 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 |
| Artturin | https://github.com/NixOS/nixpkgs/pull/218783 | 22:03:17 |
 lovesegfault | In reply to @artturin:matrix.org
You can Awesome! | 22:08:00 |
| 1 Oct 2023 |
 trofi | My understanding of strip hook is that we strip only debug symbols. Any other damage is unexpected. | 02:03:46 |
 p14 | Ping https://github.com/NixOS/nixpkgs/pull/253160 -- when I made the PR, this made it so that pkgsStatic.pkgsLLVM.clang.cc would build. But time is advancing. I would really like to land that one (or an appropriate fix) so I can unpick whatever next needs unpicking next. | 18:11:36 |
| Artturin | In reply to @p14:matrix.org
Ping https://github.com/NixOS/nixpkgs/pull/253160 -- when I made the PR, this made it so that pkgsStatic.pkgsLLVM.clang.cc would build. But time is advancing. I would really like to land that one (or an appropriate fix) so I can unpick whatever next needs unpicking next. It is surprising that we need to add the flag | 18:19:48 |
| Artturin | Is it because we bypass cmake with CFLAGS | 18:20:02 |
| Artturin | Same thing as configureFlags? | 18:20:30 |
| p14 | The issue is that without setting DCMAKE_SKIP_INSTALL_RPATH, during install time, cmake tries to rewrite the rpath in binaries which don't have any rpaths, and it fails to do so. | 18:21:31 |
| p14 | I'm not sure I've understood what you're saying about bypassing things. | 18:22:14 |
| p14 | * The issue is that without setting DCMAKE_SKIP_INSTALL_RPATH, during install time, cmake tries to rewrite the rpath in binaries which don't have any rpaths, and it fails to do so (at least, this is my understanding) | 18:24:42 |
| Artturin | Merged | 18:36:26 |
| Artturin | Cmake is dumb and requires the user to do all the things manually | 18:36:30 |
| Artturin | Doesn't even check if the binary is shared before setting the rpath | 18:37:16 |
| lovesegfault | In reply to @trofi:matrix.org
My understanding of strip hook is that we strip only debug symbols. Any other damage is unexpected. Yeah, I couldn't validate my hypothesis | 19:57:59 |
| Artturin | Weijia: why did you add loongarch64 to the bootstrap tools cross when it's not used in nixpkgs https://github.com/nixos/nixpkgs/commit/e2adc3a3a23d8735937e23105741763b05140812 | 20:52:27 |
| Artturin | asking because https://github.com/NixOS/nixpkgs/pull/258476 | 20:55:16 |
| Artturin | * asking because .https://github.com/NixOS/nixpkgs/pull/258476 | 20:55:52 |
| Artturin | * asking because https://github.com/NixOS/nixpkgs/pull/258476 | 20:56:01 |
| Artturin | Cleaning up the bootstrap to make automating the updating of the bootstrap files easier | 20:56:46 |
 Weijia | There was an effort to bootstrap a loongarch64-linux native stdenv a few months ago | 21:05:46 |
| Weijia | I assume the interest was lost after #238154, see #252590. | 21:05:49 |
| 2 Oct 2023 |
| trofi | Yeah, I personally had to give up on nixpkgs for cross-compiler infra. | 06:43:26 |
| Artturin | https://github.com/NixOS/nixpkgs/pull/258540 | 07:18:00 |
| Artturin | Preliminary work, I don't yet have a updater script but I think having the job names match the bootstrap file names will make the updater script more straightforward and have less special-cases | 07:23:26 |
| Artturin | * Preliminary work, I don't yet have a updater script but I think having the job names match the bootstrap filenames will make the updater script more straightforward and have less special-cases | 07:23:51 |
| 4 Oct 2023 |
| trofi | __FILE__ mangling is still up for review: https://github.com/NixOS/nixpkgs/pull/255192 . I did not see any regressions for the past week of trying to use it. | 05:17:50 |
| 8 Oct 2023 |
| Artturin | In reply to @trofi:matrix.org
__FILE__ mangling is still up for review: https://github.com/NixOS/nixpkgs/pull/255192 . I did not see any regressions for the past week of trying to use it. merged | 01:32:04 |
