Intermittent failure when running Postgres tests: undefined method 'name' in clean_versions!

[bronson]

Got this in two iterations while running DB=postgres rake tests.

(re-running, this appears to be a lot more rare than it would originally seem... I'm now at 50 iterations without having hit this again. First time luck.)

PaperTrailCleanerTest#test_: `clean_versions!` method No options provided should removes the earliest version(s). :
NoMethodError: undefined method `name' for nil:NilClass
    /Users/bronson/paper_trail/test/unit/cleaner_test.rb:34:in `block (4 levels) in <class:PaperTrailCleanerTest>'
    /Users/bronson/paper_trail/test/unit/cleaner_test.rb:34:in `map'
    /Users/bronson/paper_trail/test/unit/cleaner_test.rb:34:in `block (3 levels) in <class:PaperTrailCleanerTest>'
    shoulda-context (1.2.1) lib/shoulda/context/context.rb:413:in `instance_exec'
    shoulda-context (1.2.1) lib/shoulda/context/context.rb:413:in `block in create_test_from_should_hash'
    minitest (5.8.3) lib/minitest/test.rb:108:in `block (3 levels) in run'
    minitest (5.8.3) lib/minitest/test.rb:205:in `capture_exceptions'
    minitest (5.8.3) lib/minitest/test.rb:105:in `block (2 levels) in run'
    minitest (5.8.3) lib/minitest/test.rb:256:in `time_it'
    minitest (5.8.3) lib/minitest/test.rb:104:in `block in run'
    minitest (5.8.3) lib/minitest.rb:334:in `on_signal'
    minitest (5.8.3) lib/minitest/test.rb:276:in `with_info_handler'
    minitest (5.8.3) lib/minitest/test.rb:103:in `run'
    minitest (5.8.3) lib/minitest.rb:781:in `run_one_method'
    minitest (5.8.3) lib/minitest.rb:308:in `run_one_method'
    minitest (5.8.3) lib/minitest.rb:296:in `block (2 levels) in run'
    minitest (5.8.3) lib/minitest.rb:295:in `each'
    minitest (5.8.3) lib/minitest.rb:295:in `block in run'
    minitest (5.8.3) lib/minitest.rb:334:in `on_signal'
    minitest (5.8.3) lib/minitest.rb:321:in `with_info_handler'
    minitest (5.8.3) lib/minitest.rb:294:in `run'
    minitest (5.8.3) lib/minitest.rb:155:in `block in __run'
    minitest (5.8.3) lib/minitest.rb:155:in `map'
    minitest (5.8.3) lib/minitest.rb:155:in `__run'
    minitest (5.8.3) lib/minitest.rb:129:in `run'
    minitest (5.8.3) lib/minitest.rb:56:in `block in autorun'

Full output: https://gist.github.com/bronson/d597d03e7ca80ac82869

Originally reported in #696

[bronson]

Died after 78 iterations. Different error message but also happening in clean_versions! so it seems related:

ITERATION 78
warning: parser/current is loading parser/ruby22, which recognizes
warning: 2.2.3-compliant syntax, but you are running 2.2.2.
warning: please see https://github.com/whitequark/parser#compatibility-with-ruby-mri.
Running RuboCop...
Inspecting 31 files
...............................

31 files inspected, no offenses detected
Run options: --seed 11172

# Running:

..................F.................................................................................................................................................................................................................................................................................................

Finished in 18.820786s, 16.3649 runs/s, 31.5609 assertions/s.

  1) Failure:
PaperTrailCleanerTest#test_: `clean_versions!` method No options provided should removes the earliest version(s).  [/Users/bronson/paper_trail/test/unit/cleaner_test.rb:34]:
--- expected
+++ actual
@@ -1 +1 @@
-["Rosendo Howell", "Dasia Bahringer", "Wilton Dach"]
+["Rosendo Howell", "Queen Gulgowski", "Wilton Dach"]


308 runs, 594 assertions, 1 failures, 0 errors, 0 skips
rake aborted!
Command failed with status (1): [ruby -I"lib:lib:test"  "/Users/bronson/.rubies/ruby-2.2.2/lib/ruby/2.2.0/rake/rake_test_loader.rb" "test/**/*_test.rb" ]

[owenr]

This is due to Postgresql returning rows in arbitrary order unless specifically ordered. I fixed it in a commit on #689 since I originally thought it was due to AR5. Would you like a separate pull request to address this?

[jaredbeck]

Would you like a separate pull request to address this?

Yes, please, that'd be great!

[owenr]

Closed by #701 (and subsequent improvement from @batter)