Use latest patch level Ruby releases in CI

[deivid-rodriguez]

What was the end-user or developer problem that led to this PR?

Not much, just making sure we're on latest.

What is your fix for the problem, implemented in this PR?

Upgrade all the things.

Make sure the following tasks are checked

• Describe the problem / feature
• Write tests for features and bug fixes
• Write code to solve the problem
• Make sure you follow the current code style and write meaningful commit messages without tags

[martinemde]

Windows builds besides 3.2 failed. I'm rerunning with a rebase off master just to see if it's transient, but I'm not hopeful.

[simi]

no idea about the CI failures, but 👍 on update

[martinemde]

CI on windows is failing because etc is a default gem. Here's one of the errors:

You have already activated etc 1.3.0, but your Gemfile requires etc 999999. Since etc is a default gem, you can either remove your dependency on it or try updating to a newer version of bundler that supports etc as a default gem.

[martinemde]

Aha, I think etc needs to be added to the exclusions. I'll push an update.

[martinemde]

I don't know if the change I pushed should specifically target windows since they are the only failing runs, or if my patch is fine excluding it always. It would seem to be a no-op on any other build since they were already passing.

[deivid-rodriguez]

Thanks for this. From https://github.com/rubygems/rubygems/actions/runs/6441869064/job/17492104237 I see that 3.2.2 was also passing?

Yes, I normally prefer to limit these exclusions to the environments are versions where the feature is not working, so that we can keep testing other combinations still work fine, and so that we can track how the feature evolves eventually remove the exclusion.

[martinemde]

I see that 3.2.2 was also passing?

I tried to keep 3.2 tested. Did I mess up the conditional?

Yes, I normally prefer to limit these exclusions to the environments are versions where the feature is not working

We could constrain to only windows (not sure why it only shows on windows)

[deivid-rodriguez]

I tried to keep 3.2 tested. Did I mess up the conditional?

Oh god, I actually read your message and assumed the exact content of your extra commit without even looking at it 😅. Yeah, the current patch is exactly what I meant, sorry about that.

[deivid-rodriguez]

We could constrain to only windows (not sure why it only shows on windows)

Yep, that's my suggestion. Ruby On Windows adds some customisations through a operating_system.rb file that RubyGems loads when it's required. It could be that this file is now loading etc for some reason.

[martinemde]

Updated!

[deivid-rodriguez]

I can't approve my own PR but looks great!

[simi]

@deivid-rodriguez let's use my approval on your behalf :)

[martinemde]

Well you don't see this everyday (but I don't work with Ruby on Windows everyday):

       D:/a/rubygems/rubygems/bundler/tmp/1/gems/system/gems/bundler-2.5.0.dev/lib/bundler/injector.rb:90: [BUG] Segmentation fault
       ruby 3.2.2 (2023-03-30 revision e51014f9c0) [x64-mingw-ucrt]

[deivid-rodriguez]

Never seen that before, and indeed rare to see Ruby segfault these days.

I'd say this may have been surfaced by the recent optimizations merged (maybe #6970?), since the last build on master also run into that, and also a separate PR is having issues with that spec.