I18nCache: Fix overtaken class var RuntimeError in Ruby 3.0

[mcls]

This fixes errors like the one below in Ruby 3.0 by replacing the class variables with a singleton for caching the translations.

    RuntimeError: class variable @@translate_required_html of SimpleForm::Inputs::StringInput is overtaken by SimpleForm::Inputs::Base
      .../simple_form/lib/simple_form/i18n_cache.rb:13:in `class_variable_get'
      .../simple_form/lib/simple_form/i18n_cache.rb:13:in `get_i18n_cache'
      .../simple_form/lib/simple_form/i18n_cache.rb:8:in `i18n_cache'
      .../simple_form/lib/simple_form/components/labels.rb:9:in `translate_required_html'
      .../simple_form/lib/simple_form/components/labels.rb:71:in `required_label_text'
      .../simple_form/lib/simple_form/components/labels.rb:43:in `label_text'
      .../simple_form/lib/simple_form/components/labels.rb:35:in `label'

Related issue: #1720

[pedrofurtado]

Pinging maintainers for notification (some of them may be on vacation) 🤝 : @rafaelfranca @tegon @feliperenan @carlosantoniodasilva

[carlosantoniodasilva]

Just letting you know that I didn't forget about this, it's on my radar and I plan to work on it in the upcoming days alongside other Ruby 3 / Rails 6.1 upgrades. Thanks for the work here @mcls.

[LeMinh-ThienToan]

I upgraded to Ruby 3.0.0, Rails 6.1.1 and got same error.

Thanks for the work.

[carlosantoniodasilva]

@mcls I've been thinking some more about it, and considering dropping that cache entirely instead. It's extra complexity we've had to maintain over the years (and it's just growing a bit now), and I don't believe that cache is such a hot place with SimpleForm to be worth it the complexity.

That's where I'm leaning towards, I'll give it a couple more days. Let me know if you (or anyone else here) have any thoughts in the meantime. Thanks again!

[mcls]

Hey, thanks for checking. Sounds good to me.

I did a bit of digging and it looks like I18n also has its own cache which people can use if they find it necessary. I also think the default YAML backend is probably performant enough for most cases, given that it stores the translations in memory as hashes.

[carlosantoniodasilva]

Sounds good, thanks for digging some more there @mcls. I'll get back to it in the next few days and prepare a new release.

[carlosantoniodasilva]

Closed by 0595088, forgot to reference this one too. Thanks!

[matthieua]

@carlosantoniodasilva Awesome, thanks a lot for taking care of it! Do you think you can make a new release?

[carlosantoniodasilva]

@matthieua thank you. Yes, definitely, this is on my radar for this week (as well as responders/devise), but I need to make sure #1724 isn't a problem others will hit with SimpleForm when upgrading. (it doesn't appear to be SF specific, but more to investigate there.)

[matthieua]

@carlosantoniodasilva no problem, that's what I thought, thanks for the update and the awesome work with SF!

[pedrofurtado]

Thanks, @carlosantoniodasilva ! Happy to see a release with ruby 3 support asap 🎉 Thanks again! 🤝