Expect coverage for .simplecov when running with JRuby

[donv]

...when running code in .simplecov after SimpleCov.start has finished.

Downside is we get a skipped scenario indicated with dashes.

[donv]

JRuby tests are green in travis! 🍰

[donv]

@colszowka Good to merge?

[donv]

@colszowka @xaviershay @bf4 Anything more I need to do before merge?

[xaviershay]

Thanks for looking at this!

Could you add some context for why we expect these to be different? As written the scenarios are pretty confusing - why do we "expect" JRuby to have different results!?

The answer is likely - coz that's just how it works on JRuby and we don't care enough to fix it? If true, we should either a) document that, b) not specify the behaviour (i.e. existing example is "over specified" since we don't care about number of files, or c) exclude JRuby with a comment that this behaviour is undefined on JRuby.

(Option b makes the most sense to me, but I haven't thought about the problem a lot)

[donv]

Thanks for responding!

There is some discussion in #478 . As you suggested, we expect this because that's how it is.

I confronted the JRuby team, and we feel we are doing the right thing. The natural next step is filing an issue with https://github.com/ruby/spec with a test case to clear up the expected behavior. I can do that (hopefully with some help from you?).

Before that, the behavior does differ across Ruby implementations. Too bad ruby-head, ruby-2.3, and rbx don't have green tests for simplecov. That could shed some light. Anyway, I really want to have green simplecov tests for JRuby before going further. When the verdict from ruby/spec comes back, we can make changes where appropriate.

[bf4]

And https://github.com/colszowka/simplecov/blob/783eb971986e34d85e41b9ff96ae9df1e4d9c154/lib/simplecov/jruby_fix.rb

[xaviershay]

seems legit - can we put that comment in the spec (or at least in the commit message) so it's easy for future spelunkers to find?

[donv]

OK, started writing some specs for coverage in ruby/spec, and it will lead to changes in JRuby at least. So we can hold off this PR a little while longer.

[PragTob]

JRuby 9.1 seems to have resolved failures and runs cleanly now and supporting this through #547 I think this one can get closed as we don't need conditional tests then.

If anyone ( ping @donv ) disagrees please reopen/let me know :)

edit: still thanks everyone involved for your work! 🎉