Confusing default behavior, possibly due to default_retry_count

[crawfoal]

I just started using this gem, and I set up the following configuration:

RSpec.configure do |config|
  config.verbose_retry = true # show retry status in spec process
  config.default_retry_count = 4
  config.exceptions_to_retry = [Net::ReadTimeout] # only retry if this exception was thrown
end

I didn't tag any examples with :retry, because I had been looking at this answer on Stackoverflow, and got the impression that if I had this configuration, all examples that failed due to a Net::ReadTimeout error would be retried up to 4 times.

When I ran my test suite, my ouput included this:

Athlete logs a climb
  from the dashboard
[status message] Turning on truncation for the following example.
    for a nearby gym (PENDING: need to work out geolocation stubbing...)
[status message] Turning on truncation for the following example.
    for a recent gym (FAILED - 1)

User edits profile:
[status message] Turning on truncation for the following example.
  name can be changed (FAILED - 2)
[status message] Turning on truncation for the following example.
  user can select roles (FAILED - 3)

User signs out
[status message] Turning on truncation for the following example.
2nd Try error in ./spec/features/users/logout_spec.rb:4:
 Net::ReadTimeout 

RSpec::Retry: 2nd try ./spec/features/users/logout_spec.rb:4
  with valid session

# ... (omitted output) ...

Failures:

  1) Athlete logs a climb from the dashboard for a recent gym
     Got 0 failures and 2 other errors:

     1.1) Failure/Error: visit root_path

          Net::ReadTimeout:
            Net::ReadTimeout
          # ./spec/support/helpers/feature_helper.rb:14:in `capybara_login'
          # ./spec/features/athlete_logs_climb_spec.rb:62:in `block (3 levels) in <top (required)>'
          # ------------------
          # --- Caused by: ---
          # IO::EAGAINWaitReadable:
          #   Resource temporarily unavailable - read would block
          #   ./spec/support/helpers/feature_helper.rb:14:in `capybara_login'

     1.2) Failure/Error: raise Net::ReadTimeout

          Net::ReadTimeout:
            Net::ReadTimeout
          # /home/runner/.rbenv/versions/2.2.2/lib/ruby/2.2.0/net/protocol.rb:158:in `rescue in rbuf_fill'
          # /home/runner/.rbenv/versions/2.2.2/lib/ruby/2.2.0/net/protocol.rb:152:in `rbuf_fill'
          # /home/runner/.rbenv/versions/2.2.2/lib/ruby/2.2.0/net/protocol.rb:134:in `readuntil'

# ... (omitted output) ...

So I'm not sure whether those first examples that failed were retried 3 times, but the verbose message just wasn't printed, or whether they weren't retried at all. Based on how long those examples took to run compared to how long they normally take, I would suspect that they were each retried 3 times, but the message just wasn't output... is this intended behavior?

Also, the readme makes it sound like you have to either add the :retry tag to the example, or do something like this:

  # run retry only on features
  config.around :each, :js do |ex|
    ex.run_with_retry retry: 3
  end

in order for any retrying to occur, but it looks like if you specify a default_retry_count, this applies to all examples, not just the tagged ones.

[michaelglass]

Can you still reproduce this? Do you have suggestions for better copy in the readme? Happy to accept any PRs.

was display_try_failure_messages set to true?