I think we should instead try to determine the arch of the given file and if it is already universal skip. If both builds contain an x64 binary we should probably error out?

I considered that, and I'd be happy to make that change. However, some folks might want to ship part of their codebase in x64. For example, if they have a stray library that they can't yet build for arm64. That's why we went for a log instead of erroring out. In my experience, I've still seen a stray dylib in x64 even after building for Universal in Xcode.

Thoughts?

FWIW: it's a an easy process to verify the arch using:

lipo -verify_arch <arch_type>

@beyera Hm, reasonable point, I'd like to err on the side of highlighting these things to the user somehow though. How about this as a proposal:

• arm64 only binaries should be rejected (because you can't run arm64 on x64 hardware)
• x64 only binaries should initially be rejected but can be allow-listed by path in the options passed to this package

That way folks adding x64 only packages without realizing will still get warnings / errors but if they want to bypass that warning for reasons you've outlined above they still can?

That sounds very reasonable. Thanks for your input. I probably won’t get to it tonight, but I’ll make those changes within the next day or two. Cheers!

Hey @beyera, it would be great to get this in!

@MarshallOfSound
Could you please help to merge this PR and publish it. Maybe the improvements you mentioned can wait for the next PR

@gaodeng Please do not spam maintainers, my comment above stands and spamming will only annoy me, not make me want to help you.

If someone either updates this PR or raises a new one with the changes / modifications that were requested that PR can be considered. Until then there is no action to be taken here.

@beyera are you able to make the requested changes? This PR would unblock a lot of things for us.

Implemented in #47