Fix block being passed to __dry_initializer_initialize__

[bkuhlmann]

Describe the bug

Hello. 👋 I'm getting multiple warnings in my console when using Dry Initializer due to a block being passed to __dry_initializer_initialize__ when it shouldn't.

This is showing up because Ruby 2.4.1 issues warnings when you pass a block that isn't used. See Feature #15554.

To Reproduce

I don't have a great way, at the moment, to reproduce this since the warning seems to be indirectly triggered via Dry Schema because I'm not requiring Dry Initializer directly. Regardless, you'll want to ensure the following is enabled for your environment:

export RUBYOPT="-w -W:deprecated -W:performance -W:experimental"

I would expect the Dry Initializer test suite to be issuing these warnings too if warnings are enabled?

Expected behavior

I'd expect to not see this warning show up:

/Users/bkuhlmann/.cache/frum/versions/3.4.1/lib/ruby/gems/3.4.0/gems/dry-initializer-3.1.1/lib/dry/initializer/mixin/root.rb:11: warning: the block passed to '__dry_initializer_initialize__' defined at (eval at /Users/bkuhlmann/.cache/frum/versions/3.4.1/lib/ruby/gems/3.4.0/gems/dry-initializer-3.1.1/lib/dry/initializer/config.rb:112):2 may be ignored

My environment

• Affects my production application: YES
• Ruby version: ruby 3.4.1 (2024-12-25 revision 48d4efcb85) +YJIT +PRISM [arm64-darwin24.2.0]
• OS: macOS 15.2

[flash-gordon]

Thanks for reporting, I saw it too and I'll take a closer look today. It's either justified or not. In the later case we'll have to silence this warning with a monkey patch I'm afraid. The problem is there's no way know if a method yields a block or not if doesn't appear in the parameters. And even if it does appear it would be more costly than just ignoring the warning.

[flash-gordon]

Hmm, __dry_initializer_initialize__ doesn't yield a block so this warning seems justified. If we change the signature of Root#initialize to *, ** it'll help with referencing the correct source of the warning, I'll do this in main now.

[bkuhlmann]

Thanks! Yeah, only forwarding positional and keyword parameters resolves this issue nicely. I also applied this patch locally and ran it across multiple projects and don't see any issues. My console is clean now. Would be nice to release a patch next if you don't see any further complications.

[flash-gordon]

@bkuhlmann I pushed 3.2.0, please check it out

[bkuhlmann]

Thanks, yep, looks good. 🙇 I'll close this issue.