| 12 Feb 2026 |
 Stas | and seems it passed some ppc64 binaries in the LD_ paths I think | 09:58:56 |
| Stas | qemu-ppc64 seems to work fine, it is my host native binary | 09:59:42 |
|  Leonard-Orlando Menzel joined the room. | 10:20:22 |
| Leonard-Orlando Menzel | hi, can someone take a look at https://github.com/NixOS/nixpkgs/pull/489743 please? it does fix one issue for fractal to crosscompile but i am not sure if this is the most elegant way to solve it | 11:36:24 |
| Stas | if I disable modemmanager, it fails on gdk-pixbuf:
gdk-pixbuf> /nix/store/w74a7hgymrh13gknfm07nmbs8nnzx625-qemu-user-10.1.2/bin/qemu-ppc64: error while loading shared libraries: /nix/store/pyzyni2iaxjq6cw58ffyr70dcf1ydav7-glib-powerpc64-unknown-linux-gnuabielfv1-2.86.3/lib/libglib-2.0.so.0: ELF file data encoding not little-endian
| 12:01:58 |
 Puna | that sounds like something got linked incorrectly. built for the buildPlatform, but with hostPlatform library paths in its rpath. | 12:08:57 |
| Puna | i'm failing in pkgsCross.ppc64-elfv1.capnproto with pkgsCross.ppc64-elfv1.clangStdenv.cc being non-functional.
capnproto> -- Check for working CXX compiler: /nix/store/dw2m30ah26gl0rkxk585v85j5lh6sw8s-powerpc64-unknown-linux-gnuabielfv1-clang-wrapper-21.1.7/bin/powerpc64-unknown-linux-gnuabielfv1-clang++ - broken
[...]
capnproto> Building CXX object CMakeFiles/cmTC_abe92.dir/testCXXCompiler.cxx.o
capnproto> /nix/store/dw2m30ah26gl0rkxk585v85j5lh6sw8s-powerpc64-unknown-linux-gnuabielfv1-clang-wrapper-21.1.7/bin/powerpc64-unknown-linux-gnuabielfv1-clang++ -std=c++20 -MD -MT CMakeFiles/cmTC_abe92.dir/testCXXCompiler.cxx.o -MF CMakeFiles/cmTC_abe92.dir/testCXXCompiler.cxx.o.d -o CMakeFiles/cmTC_abe92.dir/testCXXCompiler.cxx.o -c /build/source/build/CMakeFiles/CMakeScratch/TryCompile-idO2pp/testCXXCompiler.cxx
[with -v added to clang shows its internal invocations]
capnproto> "ld" --hash-style=gnu --eh-frame-hdr -m elf64ppc
capnproto> clang++: error: unable to execute command: posix_spawn failed: No such file or directory
capnproto> clang++: error: linker command failed with exit code 1 (use -v to see invocation)
| 12:42:07 |
| Puna | i'll try to submit the grub thing from earlier when i have time (not rn) and i'll adjust the bootx stuff so cross is happier, but i don't have a clue for this one. you might have to setup either a VM or some other linux distro on the machine itself, build the image natively, and let the machine run for afew days… | 12:43:49 |
| Puna | * i'll try to submit the grub thing from earlier when i have time (not rn) and i'll adjust the bootx stuff so cross is happier, but i don't have a clue for this one. you might have to setup either a VM or some other linux distro on the machine itself, start building the image natively, and let the machine run for afew days to complete the builds… | 12:44:23 |
| Puna | (same error on nixpkgs master FWIW) | 12:45:51 |
| Puna | i think the issue is that we fix up the target triplet for clang by passing -target powerpc64-unknown-linux-gnu, but don't have a powerpc64-unknown-linux-gnu-ld in clangStdenv.cc.bintools. only a powerpc64-unknown-linux-gnuabielfv1-ld… guh | 13:29:26 |
| Puna | -fuse-ld=<full path to corresponding bintools wrapper>/bin/<original target>-ld works… anything against throwing this into the General clang support section somehow? | 13:44:39 |
| Puna | * -fuse-ld=<full path to corresponding bintools wrapper>/bin/<original target triplet>-ld works… anything against throwing this into the General clang support section somehow? | 13:44:49 |
| Puna | "if doing cross cross and <new target triplet>-ld doesn't exist, add the above flag" | 13:45:26 |
| Puna | or it might be time to look deeper into clang and figure out how to make it just accept our full triplets like GCC does, instead of requiring all this song-and-dance. but big scary code base… | 13:46:24 |
|  GrizzlT joined the room. | 20:33:44 |
| 13 Feb 2026 |
|  hoplopf joined the room. | 10:20:07 |
| Puna | so it definitely also links build-grub-mkfont with a freetype for the wrong platform when cross-compiling for other targets, but it only seems to complain when there is a difference in endianness? wack…
gcc -o build-grub-mkfont -I./include -std=gnu99 -fno-common -DGRUB_FILE=\"util/grub-mkfont.c\" -I. -I. -I. -I. -I./include -I./include -I./grub-core/lib/libgcrypt-grub/src/ -DGRUB_MKFONT=1 -DGRUB_BUILD=1 -DGRUB_UTIL=1 -DGRUB_BUILD_PROGRAM_NAME=\"build-grub-mkfont\" util/grub-mkfont.c grub-core/unidata.c grub-core/kern/emu/misc.c util/misc.c -I/nix/store/cc1p26vnf53s7rk01hazvi1x6r90rhw8-freetype-aarch64-unknown-linux-gnu-2.13.3-dev/include/freetype2 -L/nix/store/x49s38xdy4h12k0ynrfh5z6xv9qhppjb-freetype-aarch64-unknown-linux-gnu-2.13.3/lib -lfreetype
[...]
./build-grub-mkfont -o unicode.pf2 /nix/store/6d4rjzwgsh8vlr8pmzikxzlm5jrnd4lc-unifont-aarch64-unknown-linux-gnu-16.0.03/share/fonts/unifont.pcf.gz || (rm -f unicode.pf2; exit 1)
| 14:35:11 |
| Puna | https://github.com/NixOS/nixpkgs/pull/490128 | 14:49:00 |
| Stas | I hope to learn more about nix by osmosis just by being in these channels :) | 17:46:17 |
| 23 May 2021 |
|  @grahamc:nixos.org set the history visibility to "world_readable". | 12:31:03 |
| @grahamc:nixos.org changed the room name to "" from "". | 12:31:03 |
| @grahamc:nixos.org invited  Alyssa Ross. | 12:31:18 |
| Alyssa Ross joined the room. | 13:21:04 |
|  symphorien joined the room. | 13:29:22 |
|  joe joined the room. | 13:29:32 |
|  cleverca22 joined the room. | 13:29:44 |
|  andi- joined the room. | 13:36:08 |
|  sterni (he/him) joined the room. | 13:39:12 |
|  Server Stats Discoverer (traveler bot) joined the room. | 14:52:21 |
