Bug #4153: Minitest or ruby bug - wrong return code

• Status: Rejected
• Assignee: zenspider (Ryan Davis)
• Target version: -
• ruby -v: ruby 1.9.2dev (2010-08-02 revision 28830) [i686-linux]

[rupert (Robert Pankowecki)]

=begin
It all started here: https://github.com/jnicklas/capybara/issues/#issue/178
It seems that there is a bug in minitest which leads to returning wrong error code by the script (0 instead of 1) when errors occurs.

You can find examples of scripts that exits with the wrong code at:
https://github.com/jnicklas/capybara/issues/#issue/178/comment/604897
https://github.com/jnicklas/capybara/issues/#issue/178/comment/605003

Robert Pankowecki
=end

[zenspider (Ryan Davis)]

• Category set to lib
• Assignee set to zenspider (Ryan Davis)

=begin
I hesitate to say that this is a bug at all. It is just the way that ruby works w/ at_exit. They queue up and if someone exits inside, last exit code wins. I'm not sure what a good fix is for this. It is best reduced to:

require 'rubygems'
require 'minitest/autorun'

class TestExitBug < MiniTest::Unit::TestCase
def test_exit_bug
at_exit { exit 0 }

 flunk

end
end

% ruby x.rb; echo $?

Run options: --seed 60555

# Running tests:

F

Finished tests in 0.000515s, 1941.7476 tests/s, 1941.7476 assertions/s.

1) Failure:

test_exit_bug(TestExitBug) [x.rb:8]:

Epic Fail!

1 tests, 1 assertions, 1 failures, 0 errors, 0 skips

0

=end

[zenspider (Ryan Davis)]

=begin
I was mistaken. They stack, not queue. I have a tentative solution for this that I'm testing.
=end

[naruse (Yui NARUSE)]

=begin
What's current status?
If this is resolved, please close this.
=end

[zenspider (Ryan Davis)]

• Status changed from Open to Rejected

=begin

=end

[zenspider (Ryan Davis)]

=begin

On Dec 25, 2010, at 21:44 , Yui NARUSE wrote:

Issue #4153 has been updated by Yui NARUSE.

What's current status?
If this is resolved, please close this.

The current status is that this is really a bug in their code, but I'll be damned if I can convince them of this. I've added a workaround to minitest to deal with it, but they're off hassling all the other test framework authors instead of fixing their actual issue. I can close this ticket.

=end

[zenspider (Ryan Davis)]

=begin

On Dec 25, 2010, at 21:44 , Yui NARUSE wrote:

Issue #4153 has been updated by Yui NARUSE.

What's current status?
If this is resolved, please close this.

I should add... This is fixed on trunk, but still needs to be merged to 1_9_2. I filed a ticket before the release but 1.9.2p136 still ships with minitest 1.6.0 (approx march 2010). Is there anything I can do to help there?

=end