Skip to content

Conversation

@fretb
Copy link
Contributor

@fretb fretb commented Nov 5, 2025

fix: partially revert dependency constraint changes from #67

In #67, I proposed to update the dependency constraints to allow the gem
to be compatible with newer gem versions.

I now see that what I did was wrong, while no code changes were done
(meaning that the dependency constraints as they were would still work),
I made the gem conflict with working gem versions.

I updated the constraints to reflect what I was actually trying to do:
Make it compatible with newer gem versions, not break compatibility
with older working ones.

Signed-off-by: Frederik Thuysbaert [email protected]


@fretb
Copy link
Contributor Author

fretb commented Nov 5, 2025

@michaelklishin what do you think? Does it make sense? If possible, could this be released under a patch version, 3.1.1?

@fretb fretb force-pushed the feature/dependency-constraints branch from 013aaa9 to edd74d2 Compare November 5, 2025 14:28
@michaelklishin
Copy link
Member

michaelklishin commented Nov 5, 2025

I don't see a good enough reason to use faraday-follow_redirects 0.3 over 0.4, unless 0.4 requires a very new version of Faraday.

The rest are as good as #67.

Note that I am willing to quickly produce a 3.2.0 but after that, this dependency back-and-forth will stop. If you are not happy with the state of dependencies, go ahead and maintain a fork or figure out a way to override dependencies in your own Gemfile.

@fretb
Copy link
Contributor Author

fretb commented Nov 5, 2025

Thanks for your feedback!

I don't see a good enough reason to use faraday-follow_redirects 0.3 over 0.4, unless 0.4 requires a very new version of Faraday.

Yeah I just felt I wanted to break as little as possible in regards to the state before my previous PRs. I'll change it back to 0.4.

In ruby-amqp#67, I proposed to update the dependency constraints to allow the gem
to be compatible with newer gem versions.

I now see that what I did was wrong, while no code changes were done
(meaning that the dependency constraints as they were would still work),
I made the gem conflict with working gem versions.

I updated the constraints to reflect what I was actually trying to do:
  Make it compatible with newer gem versions, not break compatibility
  with older working ones.

Signed-off-by: Frederik Thuysbaert <[email protected]>
@fretb fretb force-pushed the feature/dependency-constraints branch from edd74d2 to b1ab37a Compare November 5, 2025 15:31
@michaelklishin
Copy link
Member

multi_json and addressable have very stable (and small) APIs, we can accept newer minors for them. I'll take care of that.

Thanks for contributing.

@michaelklishin michaelklishin merged commit 55500bd into ruby-amqp:main Nov 5, 2025
3 checks passed
@fretb
Copy link
Contributor Author

fretb commented Nov 5, 2025

multi_json and addressable have very stable (and small) APIs, we can accept newer minors for them. I'll take care of that.

I agree, my problem I was trying to fix initially with #67 was being able to use this gem in the embedded ruby environment of the newer Chef/Cinc client versions.

They recently upgraded Hashie to v5, which broke the earlier ~> 4.1 constraint here. I thought upgrading the twiddle-waka for all gems to newer minors would not be a problem, until I realized Chef/Cinc had multi_json v1.15.0 bundled.

I tried updating it upstream (chef/chef#15421), but realized that ~> 1.15 also supports 1.17, and thus I had potentially broken other persons means of just upgrading this gem in their projects without having to upgrade other dependencies.

So I made this PR, in an effort to be as frictionless as possible. :-)

@michaelklishin
Copy link
Member

Ah, you are right, we have a ~> without a patch version so it should support newer minors. Perfect.

@michaelklishin michaelklishin added this to the 3.2.0 milestone Nov 5, 2025
@michaelklishin
Copy link
Member

3.2.0 is now up on rubygems.org.

@fretb
Copy link
Contributor Author

fretb commented Nov 6, 2025

Awesome, thank you Michael! And apologies for the back and forth.

@fretb
Copy link
Contributor Author

fretb commented Nov 6, 2025

Snap, constrainting the faraday-follow_redirects to ~> 0.4 at least makes 3.2.0 break with the current version of Cinc I'm using (which has faraday-follow_redirects 0.3.0), meaning I can't upgrade easily.

As ~> 0.3 allows 0.4.0 as well, would you be open to one last PR reverting that constraint again? @michaelklishin

@michaelklishin
Copy link
Member

I will produce a patch release with a lower dependency.

@fretb
Copy link
Contributor Author

fretb commented Nov 7, 2025

Wonderful, thank you. Should I make the PR?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants