I suggest to either add checks for Diff::LCS::VERSION or re-target this to 4-0-dev branch. 4.0 is pretty much ready, but the release date is not defined. The fact that a sub-build's specs are failing can affect your decision.

@pirj OK, I checked in what I hope is a fix for thediff-lcs issue. But it looks like the tests for rspec-mocks may explicitly depend on the old behavior?

I've created another branch rebased against 4-0-dev if you think that's better, but unfortunately I don't have much time to devote to this right now so if there's more R&D or changes that need to be made in other modules, someone else may have to pick it up.

It's an interesting special case:

#<Double "double"> received :foo with unexpected arguments
  expected: ("some string", "some other string")
       got: ("this other string", "some string")
Diff:
@@ -1,3 +1,3 @@
-some other string
 some string
+this other string

Unrelated to your change

receive's with is order-dependent, and for:

expect(d).to receive(:foo).with("a", "b")
d.foo("c", "a")

it may be confusing to see

- "b"
+ "c"

because changing the code

- d.foo("c", "a")
+ d.foo("b", "a")

won't fix this spec.

It doesn't really matter if it's a list of single-line string arguments or a multi-line ones, or anything else.