ci(*): publish oci-tar-builder binary in the ci

[Mossaka]

closes #524

I didn't like the fact that the published artifacts has prefixed name containerd-shim- which doesn't really fit the purpose of the oci-tar-builder.

The oci-tar-builder binary is built for each Linux, Windows and Macos-x86_64 and Macos-arm64

[jsturtevant]

I was thinking #524 would be part of the release CI as well. Would we want to sign it too?

[Mossaka]

I was thinking #524 would be part of the release CI as well.

Ah yea I intend to work on another PR to add the release of the bianry as part of the release ci.

Would we want to sign it too?

sure :)

[jsturtevant]

I didn't like the fact that the published artifacts has prefixed name containerd-shim- which doesn't really fit the purpose of the oci-tar-builder.

I think I have a preference for modify the build action to handle this verses having a separate file and process. This would give the same steps for the signing too

[jprendes]

I didn't like the fact that the published artifacts has prefixed name containerd-shim- which doesn't really fit the purpose of the oci-tar-builder.

I think I have a preference for modify the build action to handle this verses having a separate file and process. This would give the same steps for the signing too

I agree with both of these comments.
I would like the build action to handle both cases. We could either:

• have a flag in the build action to conditionally add the containerd-shim- prefix, or
• just drop the prefix, and release unprefixed artifacts for the shims as well.

I prefer the first option. I would even defer that logic to the makefile, which already has similar logic.