supersandro2000 has quit [Ping timeout: 264 seconds]
supersandro2000 has joined #nix-darwin
daGrevis has quit [Ping timeout: 244 seconds]
daGrevis has joined #nix-darwin
iH8c0ff33 has quit [Ping timeout: 256 seconds]
<supersandro2000>
There is no easy way to detect gyp-mac-tool in every darwin build and include python?
iH8c0ff33 has joined #nix-darwin
iH8c0ff33 has quit [Ping timeout: 264 seconds]
iH8c0ff33 has joined #nix-darwin
supersandro2000 has quit [Ping timeout: 246 seconds]
supersandro2000 has joined #nix-darwin
eraserhd3 has quit [Quit: WeeChat 2.9]
eraserhd has joined #nix-darwin
supersandro2000 has quit [Ping timeout: 264 seconds]
supersandro2000 has joined #nix-darwin
philr_ has quit [Ping timeout: 260 seconds]
prusnak has joined #nix-darwin
<antifuchs>
hrm, on Big Sur (clean installation with commandline utils sans Xcode.app) I'm getting linker errors about the CoreFoundation library not being found (using the nixpkgs staging-next branch)
<antifuchs>
anyone else run into this? should I clean-reinstall the cmdline utils?
<{^_^}>
nmattia/naersk#126 (by antifuchs, 2 hours ago, open): macOS 11: CoreFoundation framework can't be linked on native builds
<antifuchs>
it seems a little suspicious since there are other issues where what I did to work around it ... worked, but it doesn't work in this case
<antifuchs>
also, seems weird to me that the ambient environment would be relevant... seems like a hermeticism violation /:
<LnL>
I wouldn't be surprised if xcbuild hardcodes /System stuff, why do you need that?
<LnL>
also ignore that comment, clt/xcode doesn't touch anything related to /System
<antifuchs>
mhm, yeah you're right - I believe /System is fine being hardcoded
<antifuchs>
now that I read the pathname more closely
<antifuchs>
funny enough, `cargo build`ing the project from rustup-installed rust works, but somehow a hermetic env doesn't
<LnL>
no it's not fine that it's hardcoded
<LnL>
since on 11.0 there are no libraries or stubs there
<LnL>
it's not clear to me what xcbuild is a workaround for, I suspect there's a better approach which might not result in these issues
<antifuchs>
I believe the xcbuild thing was a workaround for the `cc` crate that some native-building crates pull in; on darwin it would use the xcrun command if available and so putting it on native deps would fix things there
<antifuchs>
but I guess those builds are /not/ using the cc crate, so not sure what's happening there
<antifuchs>
maybe it's rust itself?
<LnL>
that's a bit handwavy, I don't understand the underlying issue
<{^_^}>
nmattia/naersk#116 (by onsails, 10 weeks ago, closed): Problem building -sys crates on darwin, direct frameworks linking
<LnL>
I know why that happens, but why is it a problem
<antifuchs>
oooh damn, I hadn't built with -K yet, so I have no binaries to inspect
<antifuchs>
I guess it's because of this: "Consequently, building crates that use newer versions of the cc crate fail, because xcrun is not available in pure build environments" from https://github.com/NixOS/nixpkgs/pull/97000
<{^_^}>
#97000 (by danieldk, 11 weeks ago, merged): stdenv: unset SDKROOT on Darwin
<antifuchs>
it's an issue in a long line of issues
<LnL>
aha xcbuild # The cc crate attempts to run xcbuild.
<LnL>
^ that should be if let Ok(sdk_path) = self.apple_sdk_root { cmd.args.push("-isysroot".into()); }
<antifuchs>
aaah that sounds very plausible, oof
<LnL>
the nix build environment already has everything setup so none of that stuff is needed there
<antifuchs>
so I think `cc` needing that fix jives with what I'm seeing in those reports; but I am suspicious that this isn't the fix that will work for the `log` crate... I don't see it using the `cc` crate at all.
<antifuchs>
so I think there's something else going on
<antifuchs>
I'll dig further
iH8c0ff33 has quit [Ping timeout: 240 seconds]
iH8c0ff33 has joined #nix-darwin
iH8c0ff33 has quit [Ping timeout: 256 seconds]
iH8c0ff33 has joined #nix-darwin
iH8c0ff33 has quit [Ping timeout: 240 seconds]
iH8c0ff33 has joined #nix-darwin
supersandro2000 has quit [Ping timeout: 265 seconds]
supersandro2000 has joined #nix-darwin
__monty__ has quit [Quit: leaving]
<antifuchs>
ahhh omg. reading through the compilation messages, one thing emerged: it was using rustc 1.45.2, which is fairly old (4 months?) - so now on rustc 1.47, it gets *much* further
<antifuchs>
guess the 1.45.2 binary isn't super able to cope with big sur, oof
philr_ has joined #nix-darwin
<LnL>
hmm, where did you get 1.45 from?
<antifuchs>
That was pinned in the repo I was building from - old nixpkgs rev, basically
<LnL>
aha, so a version without 11.0 support :)
iH8c0ff33 has quit [Ping timeout: 264 seconds]
<antifuchs>
Very much sounds like it!
<emily>
I did not know this: "Homebrew for Apple Silicon is expected to be installed in /opt/homebrew instead of the /usr/local you’re expecting."
<emily>
does Apple Silicon actually change anything there?
<emily>
if even Homebrew are biting the bullet maybe we really should consider `/opt/nix`...
<fiddlerwoaroof>
emily: so far /nix on Apple Silicon seems to work fine