 | Nixpkgs Stdenv | 230 Members |
| 75 Servers |
| Nixpkgs Stdenv | 230 Members |
| 75 Servers |
| Sender | Message | Time |
|---|---|---|
| 11 Oct 2023 | ||
 Rhys joined the room. | 02:25:33 | |
 vcunat | Hello, any hint how to avoid autotools dependencies when I need to patch ./configure? I created patch for configure along with the one for configure.ac. Configure phase is OK, but when build phase starts, it wants aclocal and I can't see why. | 08:33:40 |
 Artturin | In reply to @vcunat:matrix.orgI don't know what they use to detect changes but maybe just try touching both files? | 09:03:23 |
| Artturin | In reply to @vcunat:matrix.org* I don't know what they use to detect changes but maybe just try touching both files in postPatch? | 09:03:36 |
| vcunat | I did try that. | 09:03:43 |
| vcunat | But in the meantime I found that more of Makefile.in patching is needed than I was doing. (I just applied the same as Makefile.am but that does not suffice.) | 09:04:27 |
| vcunat | I suppose I'll just try doing autoreconf locally and hardcode the whole diff into nixpkgs patch :-( Hopefully that ugly thing will work. | 09:05:13 |
| Artturin | ah i assumed you already did that | 09:05:48 |
| vcunat | Patches that add .c and .h files are harder than I expected. | 09:05:58 |
| vcunat | Uh, it generates huge diffs of unrelated changes. Matching versions of all autotools with upstream is probably hard. | 09:08:52 |
| Artturin | try touching aclocal.m4 | 09:09:09 |
| Artturin | https://stackoverflow.com/questions/33278928/how-to-overcome-aclocal-1-15-is-missing-on-your-system-warning | 09:09:14 |
| Artturin | vcunat: | 09:10:21 |
| Artturin | or touching all 3 | 09:11:06 |
| vcunat | Thanks! Touching that file does resolve this issue. I'll most likely use it - I hope I'll manage to finish the whole hacky patching. | 09:19:51 |
| vcunat | Uh, I made some mistake. I don't think it succeeded. I probably omitted the patch by mistake. | 09:54:30 |
| vcunat | I don't know, I'll probably give it up for the current staging-next-23.05. In case someone wants to experiment, you can get prototype patch for nghttp2 (version without touching generated stuff): https://github.com/vcunat/nghttp2/pull/new/p/backport-cve-2023-44487 | 10:20:33 |
| 14 Oct 2023 | ||
 p14 | I've just discovered (messing around on staging) that a This results in NIX_LDFLAGS having a My conclusion is that the compiler only works if you're in a stdenv, but I feel that it should work standalone. Is that reasonable? I think this could be fixed by baking it into the cc-cflags as is done for compiler-rt here. https://github.com/NixOS/nixpkgs/blob/176015fc743583e30581c23bfda2bf848dd4d67d/pkgs/development/compilers/llvm/16/default.nix#L257 | 13:54:38 |
| p14 | Seems someone already beat me to it: https://github.com/NixOS/nixpkgs/issues/201591 / https://github.com/NixOS/nixpkgs/pull/220520 | 14:13:40 |
 trofi | yeah, i think mingw compilers have a similar problem and don't supply enough libraries to compile hello world without manual fiddling | 21:00:38 |
| 15 Oct 2023 | ||
| vcunat | BTW, I thought the normal way to change C compiler is to use overridden stdenv, here probably llvmPackages.stdenv or clangStdenv. | 05:59:49 |
| vcunat | Unfortunately I don't know a way of doing that in a shell except writing a shell.nix. | 06:00:24 |
| vcunat | * Unfortunately I don't know a way of doing that in a nix shell except writing a shell.nix. | 06:00:34 |
| trofi | Yeah, that matches my understanding. AFAIU `nix shell` does not create enough development environment (like running hooks). | 06:18:21 |
| 16 Oct 2023 | ||
| p14 | I don't see why it couldn't work though, right, I mean, we're building a compiler wrapper, that compiler wrapper should have enough within it to (trivially) function. Even if you have to make a stdenv in order to have any more interesting dependencies. | 16:49:36 |
| trofi | It could work in theory. I would even argue that cc should work without any wrappers and fancy variables like NIX_BINTOOLS_WRAPPER_TARGET_HOST_x86_64_unknown_linux_gnu=1. | 21:00:14 |
| 17 Oct 2023 | ||
| vcunat | What I wrote was mainly about my experience (over many years). I don't recall details. | 07:45:32 |
| vcunat | It's like, in practice some complex edge cases are difficult. | 07:47:27 |
| 21 Oct 2023 | ||
 Dusk joined the room. | 04:54:00 | |
| 23 Oct 2023 | ||
 Puna joined the room. | 20:52:50 | |