Fix error backtrace formatting on Ruby 3.4+

[orien]

Description

Context

Cucumber performs filtering and formatting on error backtraces before displaying them in terminal output.

When running on Ruby 3.4, the backtrace filtering and formatting is not working. This is defect is caused by the new backtrace format introduced in Ruby 3.4.

See:

• https://www.ruby-lang.org/en/news/2024/12/25/ruby-3-4-0-released/
• https://bugs.ruby-lang.org/issues/16495
• https://bugs.ruby-lang.org/issues/19117

Change

Update the Cucumber::Glue::InvokeInWorld and Cucumber::Formatter::BacktraceFilter classes to support both styles of backtraces.

Fixes #1770

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist:

Your PR is ready for review once the following checklist is
complete. You can also add some checks if you want to.

• Tests have been added for any changes to behaviour of the code
• New and existing tests are passing locally and on CI
• bundle exec rubocop reports no offenses
• RDoc comments have been updated
• CHANGELOG.md has been updated

[orien]

In Ruby 3.4 the Hash#inspect method produces a different format:

Ruby 3.4:

irb:001> { a: 1 }.inspect
=> "{a: 1}"
irb:002> { 'a' => 1 }.inspect
=> "{\"a\" => 1}"
irb:003> { 2 => 1 }.inspect
=> "{2 => 1}"

Ruby 3.3

irb:001> { a: 1 }.inspect
=> "{:a=>1}"
irb:002> { 'a' => 1 }.inspect
=> "{\"a\"=>1}"
irb:003> { 2 => 1 }.inspect
=> "{2=>1}"

Let's change the test to use to_json instead of inspect. This way the test expectations will be consistent across all versions of Ruby in the CI build matrix.

[luke-hill]

Good find. I know this is the issue mainly in the main branch with regards to errors, nice that to_json gives us a quick fix

[luke-hill]

I'm unresolving this so we know to look here for documentation purposes as/when a v10 is cut.

For now as part of this PR can you add some notes here: https://github.com/cucumber/cucumber-ruby/blob/main/upgrading_notes/10.0.0.md - if you are confident in doing so. Or if you'd prefer I can do it once this PR is merged. I'm flexible.

[orien]

I'm not sure what you're referring to specifically. This is a test/doc file. The changes here shouldn't affect downstream users.

[voxik]

Trying these fixes to make this cucumber-wire test pass. Unfortunately without success. Is it something what could be fixed on Cucumber side? Or does the cucumber-wire need medicine like this:

diff --git a/features/handle_unexpected_response.feature b/features/handle_unexpected_response.feature
index 2d7747e..4e0aa8c 100644
--- a/features/handle_unexpected_response.feature
+++ b/features/handle_unexpected_response.feature
@@ -25,7 +27,7 @@ Feature: Handle unexpected response
       | ["begin_scenario"]                                   | ["yikes"]                           |
       | ["step_matches",{"name_to_match":"we're all wired"}] | ["success",[{"id":"1", "args":[]}]] |
     When I run `cucumber -f pretty`
-    Then the output should contain:
+    Then the output should match:
       """
-      undefined method `handle_yikes'
+      undefined method [`']handle_yikes'
       """

[luke-hill]

Partially reviewed, probably won't have time this week. But this is near the top of my list.

Great work finding a good replacement!

[luke-hill]

Can we change this to be hard-coded. Aruba steps are designed to showcase simple examples and this has 3 parameter types for "I have no idea what reason...." So if this just becomes When('monkey eats banana') then we can make the logger step also hard coded and the test showcasing that the logger having 3 items gets translated will be a bit easier to understand

Notwithstanding that, I'm not sure I agree that "is" what happens though. So I'll probably need to dig into this.

[orien]

Updated in 4740677.

[luke-hill]

Fully reviewed this PR now. If you can respond to the points we can look to get this merged in soon and included in v10.

You'll also likely need to do some form of rebase. But it might be worth waiting as I'm doing a few derpy things in main currently.

[luke-hill]

Add new line and wrap expected string in ()

[orien]

Fully reviewed this PR now. If you can respond to the points we can look to get this merged in soon and included in v10.

You'll also likely need to do some form of rebase. But it might be worth waiting as I'm doing a few derpy things in main currently.

Thanks @luke-hill. I think I've addressed everything and rebased the branch against main. Please let me know if/when I need to rebase again, if you're still working for v10.

[luke-hill]

3/6 files are marked as good from me, couple of minor things to fix - I'll leave the note up for the doc updates as I still haven't got round to that yet. I will soon though as I've merged my chunky v10 diff.

[luke-hill]

For now @orien this is in a decent spot. So you can rest up for a while now as there's not much to do.