paketo-buildpacks / rfcs

Apache License 2.0
19 stars 33 forks source link

Implement RFC0009: Remove `yj` binary #293

Closed loewenstein closed 1 year ago

loewenstein commented 1 year ago

RFC

Summary

Remove the yj binary from all stacks' build images

Stack Adoption

loewenstein commented 1 year ago

Hey @paketo-buildpacks/stacks-maintainers,

here's the implementation tracking issue for the yj removal RFC.

robdimsdale commented 1 year ago

With the PRs merged above for all jammy stacks, this is complete.

sophiewigmore commented 1 year ago

@robdimsdale unfortunately, I think the stack version bumps should have been a minor bump, since we removed a package which could break folks. Do you think we should cut new minor releases to make this clear?

robdimsdale commented 1 year ago

@sophiewigmore I think they should either be a major or a patch. Depending on whether you view yj as part of the surface area of a stack or not.

I don't, hence the patch. We could could majors if you want though.

sophiewigmore commented 1 year ago

Why don't you view it as a part of the surface area of the stack? I think it would be if it was in the stack before! Also, I don't think any of our stacks are at a 1.0.0 release yet, which would make a minor bump the best option IMO

robdimsdale commented 1 year ago

Maybe it would be more accurate to say I don't consider the yj library part of the supported surface area of the stack.

My understanding is that yj was added to help simple shell-based buildpacks interact with TOML files. I don't think anyone actually uses it for that feature as all of our buildpacks have TOML parsing libraries in golang. I view its existence as a bug, not a feature. So I view its removal as a bugfix, not a breaking change.

Also, pre v1.0.0 semver doesn't really exist, so I interpret that to mean we have flexibility to choose our own versions.

Personally I don't think the majority of Paketo consumers ever look at stack versions because they're always hidden inside a builder.

All of that being said, I don't really feel strongly about this. If you think a minor bump is more clear to consumers, I have no objections to you re-releasing all the stacks with a minor bump.