Cross compile macos

[hulto]

What type of PR is this?

/kind feature

What this PR does / why we need it:

• Adds support to cross compile Imix to MacOS from the devcontainer
• Updates get_ip dependency to something more stable

Which issue(s) this PR fixes:

[Cictrone]

Is the lack of cidr/mac a regression?

[hulto]

@claude How's this?

[claude]

Claude finished @hulto's task —— View job

---

PR Review

• Read repository guidelines (CLAUDE.md)
• Examine PR changes and diff
• Review code changes
• Address review comment about cidr/mac regression
• Provide comprehensive feedback

---

Summary

This PR successfully adds cross-compilation support for MacOS from the devcontainer and simplifies the network interface handling. However, there are critical issues that need to be addressed before merging.

Critical Issues

1. build.rs has compilation errors 🔴

implants/build.rs:15 references undefined variables:

println!("cargo:rustc-cfg=host_os=\"{}\"", HOST_FAMILY);

Should be HOST_OS (defined on lines 2-11), not HOST_FAMILY. Also missing the anyhow::Result import:

use anyhow::Result;

fn main() -> Result<()> {
    println!("cargo:rustc-cfg=host_os=\"{}\"", HOST_OS);  // Fix: HOST_OS not HOST_FAMILY
    Ok(())
}

2. Answering @Cictrone's review question about cidr/mac regression ⚠️

Yes, this is a regression - the PR removes important functionality:

Before (network-interface crate):

• ✅ CIDR notation (e.g., 172.17.0.2/24)
• ✅ MAC addresses (e.g., 02:42:ac:11:00:02)
• ✅ Multiple IPs per interface in single entry
• Return format: { name, ips: [], mac }

After (local-ip-address crate):

• ❌ No CIDR notation (just 10.0.124.42)
• ❌ No MAC addresses
• ❌ Separate entries per IP address on same interface
• Return format: { name, ip }

Impact:

• The arp_scan tome was deleted (tavern/tomes/arp_scan/) - possibly because MAC addresses are no longer available
• Breaking API change for any code/scripts relying on sys.get_ip() structure
• Loss of network mask information useful for subnet calculations

Documentation & DevContainer

Positives ✅

1. Excellent documentation in docs/_docs/user-guide/imix.md with clear MacOS cross-compilation setup instructions
2. Well-structured devcontainer setup with Zig for cross-compilation
3. Proper SDK mounting instructions for Rancher desktop limitations
4. Updated return format examples in docs/_docs/user-guide/eldritch.md:908

Code Quality

get_ip_impl.rs changes

Positives:

• ✅ Significantly simpler (83 fewer lines)
• ✅ Removed complex netmask-to-CIDR conversion logic
• ✅ Test updated appropriately

Concerns:

• ⚠️ Error handling degraded: .unwrap() on line 22 will panic instead of returning Result
• ⚠️ Loss of functionality may impact operational use cases

Dependency change

• network-interface → local-ip-address: Simpler but more limited functionality
• May be "more stable" as mentioned in PR description, but trades features for stability

Recommendations

1. Fix build.rs immediately - won't compile as-is
2. Consider the breaking change implications:
   • Is the loss of CIDR/MAC info acceptable for your use cases?
   • Should this be a major version bump?
   • Are there other tomes/scripts using the old format?
3. Improve error handling: Replace .unwrap() in get_ip_impl.rs:22 with proper error handling
4. Consider: If MAC addresses and CIDR are needed, might need to keep network-interface or find alternative

Questions for PR Author

1. Was the removal of CIDR/MAC intentional to support MacOS cross-compilation, or a side effect of the library change?
2. Are there known use cases that depend on MAC addresses or CIDR notation?
3. Has this been tested with actual MacOS cross-compilation in the devcontainer?

---

[Cictrone]

sus as hell