Phil Pirozhkov

Please accept my apologies. That must have been `before` inside `before`, not `let`. Alright, it's an edge case, but worth and possible to fix. I suggest adding a check that...

Quite an interesting case, @luke-hill Wondering how ‘source do’ works? In RuboCop’s own specs we use heredocs. But I have no certainty that Cucumber actually parses those, probably just evals...

Sorry for the noise from transferring back and forth, it slipped my mind some Capybara cops remain here. Speaking of the issue itself. We have a check ‘return unless inside_example_group?(node)’,...

Is this an issue from the real code? Why would anyone want to use that?

If those are typos, surely let’s add a cop to fix those

‘change’ is a Rails thing? And it’s probably a good recommendation for Rails code in general, not just specs? Would ‘round’ be always identical to ‘change’ (i mean floor/ceil)? Good...

I see, thanks for this research. With what you say, it seems that indeed, this cop makes more sense in Rails RSpec specs. I recall that the the default timestamp...

user.created_at - floating point, 6 to nine digits of precision tTime.now.utc.round - integer, zero precision Chances of ‘eq’ to match are one in a million. Why do we even need...

Could ‘change’ be considered as an option to write time comparisons more reliably? I suggest starting from scratch, getting examples of unreliable specs, and finding a common solution we could...

Pending reason and the docstring are different.