| 26 May 2021 |
andi- | Ahh, there is a pending commit that I have to push. Sorry for that. | 12:01:11 |
andi- | Pull again then it should also work on darwin (Clang Stdenvs) | 12:02:33 |
thefloweringash | andi-: I haven't looked at LTO on macos, but from tickets like https://github.com/NixOS/nixpkgs/issues/19098 I just thought it didn't work. Did that change? | 12:10:39 |
andi- | I don't know! I am trying to figure that out :-) | 12:11:05 |
andi- | but it keeps crashing on llvm :/ | 12:11:13 |
sterni (he/him) | thefloweringash: if there is no linker plugin it may not work at all yeah | 13:02:20 |
sterni (he/him) | I wonder if we could use lld for the non bootstrap python derivation though? | 13:02:44 |
sterni (he/him) | andi-: it fails at clang-7: error: unknown argument: '-fno-semantic-interposition' currently for minimal python on nixos currently | 14:58:45 |
andi- | mhm, perhaps the clang version is too old? I know that python 3.10 enables this by default with --enable-optimsations and only recently started doing so on GCC as well. | 15:00:49 |
Finn Behrens | In reply to @andi:kack.it Ahh, there is a pending commit that I have to push. Sorry for that. .#python did build (as arm64). readline does not seem to work, so tab is not working in the python shell | 15:00:52 |
andi- | no idea if readline is supposed to work on the python build on MacOS. It seems to work on the old revision on my device 🤔 | 15:02:11 |
andi- | I am currently at 99% of clangd. So far the device hasn't crashed again... | 15:02:32 |
Finn Behrens | I also have 11.5, maybe this helped to not let it crash | 15:03:03 |
Finn Behrens | * I also have 11.5 (20G5023d), maybe this helped to not let it crash | 15:04:12 |
Finn Behrens | In reply to @andi:kack.it no idea if readline is supposed to work on the python build on MacOS. It seems to work on the old revision on my device 🤔 It does not even work at the python version under /usr/bin/python. so probably expected behavior for python2 | 15:06:38 |
andi- | Can you try python3? | 15:07:06 |
Finn Behrens | have to build it. but on it | 15:07:13 |
andi- | Given that Python2 worked that means LTO did actually work?!? | 15:07:54 |
Finn Behrens | Btw, not sure if you saw this, but I'm trying it with flakes. so I am building lecagyPackages.aarch64-darwin.python. also my nix is x86 as lowdown did not worked on my system so far | 15:08:18 |
andi- | hah, my system is similar :D I have an overlay that pulls in all haskell related stuff from the x86_64-darwin equivalent as otherwise pandoc etc.. would not work. | 15:09:11 |
Finn Behrens | In reply to @andi:kack.it hah, my system is similar :D I have an overlay that pulls in all haskell related stuff from the x86_64-darwin equivalent as otherwise pandoc etc.. would not work. nice, did not do that yet | 15:09:28 |
Finn Behrens | .#python3 is not building | 15:09:37 |
Finn Behrens | Download python3-aarch64-darwin.log | 15:10:04 |
sterni (he/him) | Finn Behrens: what is the lowdown failure, I'm curious | 15:10:35 |
sterni (he/him) | also you could always downgrade to nix stable :p | 15:10:48 |
Finn Behrens | someting involving sandboxing. this is mainly happening, when building github:nixos/nix#packages.aarch64-darwin.nix | 15:11:30 |
andi- | # nested sandboxing doesn't seem to work so lets disable the feature in the build
lowdown = super.lowdown.overrideAttrs (_: {
postConfigure = ''
for file in *.c; do
sed -e s/HAVE_SANDBOX/HAVE_NO_SANDBOX_SHIT/g -i $file
done
'';
});
| 15:11:44 |
andi- | I have that in my macbook specific config | 15:11:49 |
Finn Behrens | In reply to @sternenseemann:systemli.org also you could always downgrade to nix stable :p but then I can't load my flakes. And I'm not able to manage my nix path (to layzie xD) | 15:11:59 |
Finn Behrens | In reply to @andi:kack.it
# nested sandboxing doesn't seem to work so lets disable the feature in the build
lowdown = super.lowdown.overrideAttrs (_: {
postConfigure = ''
for file in *.c; do
sed -e s/HAVE_SANDBOX/HAVE_NO_SANDBOX_SHIT/g -i $file
done
'';
});
nice, just have to get it into nix, as the use an own version of lowdown :-) | 15:12:27 |