| 5 Jul 2021 |
 Alyssa Ross | hmm, I thought it did when I checked | 14:21:08 |
 sterni (he/him) | Alyssa Ross: the non cross compiled one does | 14:22:24 |
| sterni (he/him) | ldd ./result-dev/bin/llvm-config
linux-vdso.so.1 (0x00007ffc18a91000)
libLLVM-11.so => /nix/store/52szvdc5qkh7llzpwamxjwg2cldz7m46-llvm-11.1.0-lib/lib/libLLVM-11.so (0x00007ff3bb43a000)
libpthread.so.0 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/libpthread.so.0 (0x00007ff3bb419000)
libstdc++.so.6 => /nix/store/d9mfvhvksvsx0ar8pmxwhgwd9my045rp-gcc-10.3.0-lib/lib/libstdc++.so.6 (0x00007ff3bb244000)
libm.so.6 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/libm.so.6 (0x00007ff3bb101000)
libgcc_s.so.1 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/libgcc_s.so.1 (0x00007ff3bb0e7000)
libc.so.6 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/libc.so.6 (0x00007ff3baf24000)
libffi.so.7 => /nix/store/qv3b7pswf4j9c03dchr96p4r27700ph5-libffi-3.3/lib/libffi.so.7 (0x00007ff3baf17000)
libz.so.1 => /nix/store/0b0mijbd7mp8gcglmkkdr2775ki38x9v-zlib-1.2.11/lib/libz.so.1 (0x00007ff3baefa000)
librt.so.1 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/librt.so.1 (0x00007ff3baeef000)
libdl.so.2 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/libdl.so.2 (0x00007ff3baee8000)
libncursesw.so.6 => /nix/store/sfnp4faf4m8d8p04jvmabcn7j5bacqqg-ncurses-6.2/lib/libncursesw.so.6 (0x00007ff3bae76000)
libxml2.so.2 => /nix/store/j9af3y4n3viladi76xmpnaz7cps68hcd-libxml2-2.9.12/lib/libxml2.so.2 (0x00007ff3bad08000)
/nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib/ld-linux-x86-64.so.2 => /nix/store/adxc893j47gxx3xjw403zdf0liiddvw2-glibc-2.32-48/lib64/ld-linux-x86-64.so.2 (0x00007ff3c0f8b000
| 14:22:40 |
| sterni (he/him) | (this is pkgs.llvmPackages_11.llvm.dev) | 14:22:53 |
| Alyssa Ross | hmm, I see | 14:22:54 |
| sterni (he/him) | maybe its optional and it doesn't find it for the cross case and ignores it | 14:23:16 |
| sterni (he/him) | and in llvm 12 it does discover it, but the wrong one, possibly? | 14:23:26 |
 ius | Pff, finally managed to override patchelf the way I wanted (i.e. for cross only): I ended up prepending it to extraNativeBuildInputs in stdenv/cross. Original patchelf is still dragged in somehow, but ends up further down in PATH | 17:55:06 |
|  spacesbot - keeps a log of public NixOS channels joined the room. | 19:21:38 |
| spacesbot - keeps a log of public NixOS channels | 19:49:35 |
| 6 Jul 2021 |
|  nurelin joined the room. | 11:43:42 |
|  7c6f434c left the room. | 15:50:41 |
| spacesbot - keeps a log of public NixOS channels changed their display name from spacesbot to spacesbot - keeps a log of public NixOS channels. | 22:11:42 |
| 9 Jul 2021 |
|  vika (she/her) 🏳️⚧️ set a profile picture. | 16:38:59 |
| 11 Jul 2021 |
| sterni (he/him) | Redacted or Malformed Event | 11:30:34 |
| sterni (he/him) | Redacted or Malformed Event | 12:44:59 |
| ius | When crossSystem is set when importing nixpkgs, aren't all entries outside of buildPackages supposed to target the cross architecture? | 13:32:13 |
| ius | "powerpc-unknown-linux-musl"
nix-repl> hello
«derivation /nix/store/6axas7kiammp06swr9kr8582jm82lyag-hello-powerpc-unknown-linux-musl-2.10.drv»
nix-repl> getent
«derivation /nix/store/j97rd009di0d19z592zn6qvvmzmbw9ks-getent-1003.1-2008.drv»```
| 13:32:28 |
| ius | * "powerpc-unknown-linux-musl"
nix-repl > hello
«derivation /nix/store/6axas7kiammp06swr9kr8582jm82lyag-hello-powerpc-unknown-linux-musl-2.10.drv»
nix-repl > getent
«derivation /nix/store/j97rd009di0d19z592zn6qvvmzmbw9ks-getent-1003.1-2008.drv»```
| 13:32:54 |
| ius | * nix-repl > stdenv.targetPlatform.config
"powerpc-unknown-linux-musl"
nix-repl > hello
«derivation /nix/store/6axas7kiammp06swr9kr8582jm82lyag-hello-powerpc-unknown-linux-musl-2.10.drv»
nix-repl > getent
«derivation /nix/store/j97rd009di0d19z592zn6qvvmzmbw9ks-getent-1003.1-2008.drv»```
| 13:33:34 |
| sterni (he/him) | ius: getent likely only sets name and not pname / version, so the config isn't mangled into the final derivation name | 13:33:58 |
| ius | hmm, so you're saying it should target ppc, yet just doesn't have a mangled name? | 13:34:46 |
| ius | Cause I ran into a build failure of one of its dependencies, and ended up being confused whether it should be a cross package or not. | 13:35:54 |
| sterni (he/him) | In reply to @ius:nltrix.net
hmm, so you're saying it should target ppc, yet just doesn't have a mangled name? yes | 13:36:30 |
| sterni (he/him) | getent is a wrapper derivation generated using runCommand, so it doesn't have the pname/version split | 13:37:06 |
| ius | Right, if I build it separately I can see it builds a derivation of the mangled version as well as the wrapper. In that case, the issue is in the dependency | 13:38:13 |
| ius | ar x /nix/store/y7z59i6l25m46jdqgqxjpbzsq4npd9mb-compat-netbsd-9.1-powerpc-unknown-linux-musl/lib/libnbcompat.a db.lo
# file db.lo
db.lo: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), not stripped
| 13:42:45 |
| ius | Git history tells me it should cross compile, but I don't see how that's supposed to work looking at the source: https://github.com/NixOS/nixpkgs/blob/master/pkgs/os-specific/bsd/netbsd/default.nix#L160
# the build system re-runs `./configure` with `HOST_CC` (which is their
# name for Build CC) as a compiler to make `defs.mk`, which is installed
depsBuildBuild = [ buildPackages.stdenv.cc ] ++ commonDeps;
HOST_CC = "${buildPackages.stdenv.cc.targetPrefix}cc";
| 13:48:40 |
| ius | It seems to run configure with a correct --host, but then run it again with CC set to the build cc, then make afterwards obviously results in wrong output | 13:51:03 |
| 12 Jul 2021 |
|  Rick (Mindavi) joined the room. | 08:17:49 |
