Fix builds for Rails 7.1.0

[yuki24]

Right now, Ransack fails to execute on Rails 7.1.0beta1 with the error below:

  1) Ransack::Search#method_missing allows chaining to access nested conditions
     Failure/Error: table        = attr.arel_attribute.relation.table_name

     NoMethodError:
       undefined method `table_name' for #<Arel::Table:...>

This is because the table_name method has been dropped as part of this commit rails/rails@1d98bc5, which has made Rails imcompatible with Ransack.

This PR adds a check for the existence of the table_name method to restore the compatibility.

[jlupox]

Thanks @yuki24.
It doesn't work for me. I have to change to:

table = relation.respond_to?(:table_name) ? relation.try(:table_name) : relation.name

[yuki24]

@jlupox Thanks for pointing that out! It was embarrassingly broken 😅 But it should be fixed now.

[engineersmnky]

I would recommend just changing this to

table        = attr.arel_attribute.relation.name

Checking for the existence of table_name is unnecessary overhead as Arel::Table#name has always existed.

[yuki24]

@engineersmnky That didn't work, as it's not always an Arel::Table object but could be an Arel::TableAlias object, in which case there are 19 failing tests with the error below:

  18) Ransack::Search#method_missing allows chaining to access nested conditions
      Failure/Error: raise "No table named #{table} exists."

      RuntimeError:
        No table named children_people exists.
      # ./lib/ransack/adapters/active_record/context.rb:20:in `type_for'
      # ./lib/ransack/nodes/attribute.rb:35:in `type'
      # ./lib/ransack/nodes/condition.rb:266:in `default_type'
      # ./lib/ransack/nodes/condition.rb:25:in `extract'
      # ./lib/ransack/nodes/grouping.rb:175:in `write_attribute'
      # ./lib/ransack/nodes/grouping.rb:155:in `block in build'
      # ./lib/ransack/nodes/grouping.rb:150:in `each'
      # ./lib/ransack/nodes/grouping.rb:150:in `build'
      # ./lib/ransack/nodes/grouping.rb:146:in `new_grouping'
      # ./lib/ransack/nodes/grouping.rb:95:in `block in groupings='
      # ./lib/ransack/nodes/grouping.rb:94:in `each'
      # ./lib/ransack/nodes/grouping.rb:94:in `groupings='
      # ./lib/ransack/search.rb:106:in `method_missing'
      # ./spec/ransack/search_spec.rb:687:in `block (3 levels) in <module:Ransack>'

[yuki24]

@engineersmnky The try call is not there any more. It would have been greatly appreciated if you had read the actual code in the PR.

[yuki24]

Rails 7.1 release is right around the corner, and it would be great if we could get this merged. The tests are all green and the Code Climate check seems trivial (only 1 point of complexity check).

[jayshepherd]

Rails 7.1 release is right around the corner, and it would be great if we could get this merged. The tests are all green and the Code Climate check seems trivial (only 1 point of complexity check).

Looks like it was this morning.. Mind bumping to 7.1?

[yuki24]

@jayshepherd The world moves too fast! Just updated the CI config and this should be good now.

[davidwessman]

Is there something we can help with to move this along? Testing?

[n-rodriguez]

Hi there! Any news?

[deivid-rodriguez]

Hello! I'll try to merge and get a release out this week, thanks for the patience!

[deivid-rodriguez]

Looks great, thanks!

[deivid-rodriguez]

Hei, I was expecting to release this week as I said, but I realized I'd like to give @scarroll32 a chance to validate what's going in, and leave whatever feedback he has. So I'd let this PR sit for a couple of days in main before releasing.

Apologies if I created some false expectations there! New ETA around the middle of next week unless Sean gets to it earlier.

@scarroll32 I left a release draft ready in case you want to have a look. There's not much to be released, really.

[scarroll32]

Looks good to me, thank you @deivid-rodriguez and @yuki24

[deivid-rodriguez]

Awesome, I will release now then. By the way, I sneaked #1447 into the release in the last minute and will include it too!

[iainbeeston]

Could this be back ported to ransack 3? I'm stuck with ransack 3 until #1403 is resolved

[Augustin-Grenne]

I'd like to second @iainbeeston request to backport this to ransack 3 as sadly patching 4 proofed insufficient (see my comment) and I'm also waiting on #1403