| 21 Jan 2026 |
 Yureka (she/her) | or was that failure only at runtime? | 17:19:38 |
 hexa | * onnxruntime will regress firefox fwiw š | 17:21:35 |
| hexa | the python one happens on import, so yeah at runtime | 17:21:59 |
| hexa |
Check whether the following modules can be imported: onnxruntime
ImportError: /nix/store/n1xc96lcmwg4dwdh9akx3yycsp153rij-python3.13-onnxruntime-1.23.2/lib/python3.13/site-packages/onnxruntime/capi/onnxruntime_pybind11_state.so: cannot enable executable stack as shared object requires: Invalid argument
| 17:22:13 |
 dotlambda | https://github.com/NixOS/nixpkgs/pull/482363 | 18:01:52 |
 GaƩtan Lepage | Weird, it works fine on master... Could it be a pybind11 change causing this build failure? | 21:32:28 |
 leona | it's the glibc change | 21:33:16 |
| leona |
dlopen and dlmopen no longer make the stack executable if a shared library requires it, either implicitly because of a missing GNU_STACK ELF header (and default ABI permission having the executable bit set) or explicitly because of the executable bit in GNU_STACK, and the stack is not already executable. Instead, loading such objects will fail.
| 21:34:01 |
| GaƩtan Lepage | Oh right. So the "proper fix" that was supposedly integrated in 1.23.1, is not really enough? Should we add these flags back then? | 21:38:49 |
| leona | Apparently or it's something in the pybind foo in the python library, I think nobody here knows | 21:39:26 |
| GaƩtan Lepage | Reverting https://github.com/NixOS/nixpkgs/pull/479191 fixes python3Packages.onnxruntime. | 21:45:48 |
| GaƩtan Lepage | Setting env.NIX_LDFLAGS = "-z,noexecstack"; in the python derivation doesn't seem to be necessary. | 21:48:40 |
| hexa | well, those are linker flags | 21:48:59 |
| hexa | they are applied when the library gets linked | 21:49:18 |
| hexa | * they are applied when the library gets linked together from its individual objects | 21:49:31 |
 mdaniels5757 | I was asked if https://github.com/NixOS/nixpkgs/pull/482131 can target staging-next. 543 rebuilds on Linux, 8k on Darwin. | 21:54:33 |
| GaƩtan Lepage | Opened https://github.com/NixOS/nixpkgs/pull/482435 | 21:55:10 |
| mdaniels5757 | * I was asked if https://github.com/NixOS/nixpkgs/pull/482131 can target staging-next. 543 rebuilds on Linux, 8k on Darwin. Thoughts? | 21:55:31 |
 FabiƔn Heredia | I would say no due to darwin, the current staging-next cycle is mostly only waiting on darwin and doesn't seem to have any urgent changes (eg, security or critical bug) | 21:56:53 |
 emily | why does less rebuild so much more on Darwin? | 22:31:03 |
 r-burns | maybe because of this? https://github.com/NixOS/nixpkgs/blob/7fd1c9c7b53ed52f49adb2034263397adc55da17/pkgs/os-specific/darwin/apple-source-releases/file_cmds/package.nix#L122 | 22:45:19 |
| 22 Jan 2026 |
| emily | that would do it | 00:05:09 |
| emily | we could probably leave that late-bound | 00:05:24 |
| emily | especially since it looks like we'd currently be ignoring the user's preference? | 00:05:39 |
| emily | in fact the current syntax looks pretty busted :)
if [ `basename $0` = "zless" ] ; then
pager=${/nix/store/56f32dkj830jnh5r3bihqiys1cz6j4ag-less-679/bin/less}
else
pager=${/nix/store/56f32dkj830jnh5r3bihqiys1cz6j4ag-less-679/bin/more}
fi
| 00:06:11 |
| emily | PRs welcome to just drop that substitution | 00:06:22 |
 VladimĆr ÄunĆ”t | No blockers for merging staging-next soon? | 12:42:43 |
 K900 | Not that I know of | 13:03:48 |
| K900 | https://blog.rust-lang.org/2026/01/22/Rust-1.93.0/ | 14:04:56 |
| K900 | Looks like this one will explode | 14:05:20 |
