supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nix-darwin
supersandro2000 has quit [Quit: The Lounge - https://thelounge.chat]
supersandro2000 has joined #nix-darwin
stephank has quit [Quit: stephank]
stephank has joined #nix-darwin
supersandro2000 has quit [Quit: The Lounge - https://thelounge.chat]
supersandro2000 has joined #nix-darwin
supersandro2000 has quit [Client Quit]
supersandro2000 has joined #nix-darwin
aminechikhaoui has quit [Quit: The Lounge - https://thelounge.github.io]
aminechikhaoui has joined #nix-darwin
philr has quit [Ping timeout: 240 seconds]
ece8 has joined #nix-darwin
ece has quit [Read error: Connection reset by peer]
ece8 is now known as ece
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nix-darwin
ece has quit [Ping timeout: 268 seconds]
__monty__ has joined #nix-darwin
<__monty__> LnL: Am I on the right trail with the python3-minimal build failure? Last section in the gist, https://git.io/JmxTi
<LnL> definitively sounds plausible, python does lots of dynamic loading stuff
<__monty__> LnL: I'm trying to repeat the conftest, removing -I's and -L's from the clang command to figure out where it's finding futimens but it compiles fine without any of them >.< Can I simply reproduce the environment the actual configure runs in using nix-shell -p stdenv.__bootPackages.stdenv...?
<LnL> the libc flags are part of the stdenv so you can't really get rid of it once the wrappers are built
<LnL> to get the correct build environment you want -A
<LnL> using -p is equivalent to -E 'import <nixpkgs> {}; stdenv.mkDervation { buildInputs = [ .. ]; }' so it's always the top level stdenv
<__monty__> Is there a better way to figure out why python3-minimal isn't building/why the build is complaining about implicit function declarations?
<LnL> is this with or without the patched stubs?
<__monty__> With the patched stubs.
<LnL> I'm almost certain that's the issue then, so unless deployment target helps there somehow the headers and stubs will need to be in sync
<__monty__> darwin-stubs and the SDK? They should be though.
<LnL> you changed the stubs but the headers are still from 10.12
<__monty__> Where are they coming from, apple_source_releases?
<LnL> Libsystem is a bit of a mix of everything but yet should mostly be from the source releases
<__monty__> So I have to find all the 10.13 source releases? What do I do if some aren't open anymore? I think recent OS versions have fewer open source releases.
<__monty__> Oh, right. I'm mixing things up. It's recent SDK versions that seem to be harder to find.
<LnL> most of those should be in there, some source releases have indeed stopped at some point but unless things moved somehwere else using the old version usually mostly works
<LnL> this specifically probably comes from xnu, which is one of the core parts like Libc for which the headers will always be available somewhere otherwise you can't build anything
<LnL> and if I'm not mistaken we're not using the bootstrap Libsystem at this point anymore so that doesn't have to be touched yet
<__monty__> LnL: So I add an osx-10.13.6 version which is a copy of osx-10.12.6 except for the xnu version?
<__monty__> Oh, or it can just have an xnu attr I guess.
<LnL> the source releases are kind of namespaced by os version
<LnL> the "osx-10.12.6" determines where in comes from in the versions set
<LnL> that's mostly just for reference tho
<LnL> the end result is still darwin.xnu so nothing changes outside of that file
<__monty__> And you're saying I should only update xnu? Or is that for now until I run into more problems and gradually try to update everything to the 10.13.6 source releases?
<__monty__> Or maybe just the ones with a bullet on the opensource.apple.com page?
<LnL> up to you, you can start with that but I'd definitively try to bump all of those as a minimum along with the sdk
<LnL> not everything in there is as important, if you take a look at the versions you can see that there's even a bunch of stuff that's still on 10.11
<__monty__> LnL: How do I distinguish what should be in the attrset? Is it limited to things that appear in the generated macos-10.13.6.nix? Or does DiskArbitration belong in it too? (The first project with a bullet on the opensource page.)
<LnL> I don't follow
<__monty__> xnu is on the opensource.apple.com/10.13.6 page with a bullet. It is also in the nix file generated using `generate-sdk-packages.sh` and it appears in the various attrsets in apple_source_releases/default.nix.
<__monty__> DiskArbitration appears with a bullet but isn't in either of the files.
<LnL> ah, we don't package everything from opensource.apple.com
<LnL> everything in that list was added for some reason, like headers for Libsystem
<LnL> no idea what that generate script is for, it's new
<__monty__> LnL: Hmm, so futimens should indeed come from xnu. But this python3-minimal is being built as part of its dependencies. So bumping xnu doesn't and can't solve the problem 🤔
<eraserhd> During darwin-rebuild, what do X, Y, and Z mean in [X/Y/Z built, ...?
<eraserhd> running, completed, total?
eraserhd2 has joined #nix-darwin
eraserhd has quit [Ping timeout: 240 seconds]
eraserhd2 has quit [Ping timeout: 240 seconds]
eraserhd2 has joined #nix-darwin
eraserhd2 is now known as eraserhd
philr has joined #nix-darwin
__monty__ has quit [Quit: leaving]
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nix-darwin