Specializations of assertEquals() and assertNotEquals()

[sebastianbergmann]

Implement the following methods

• assertEqualsCanonicalizing()
• assertEqualsIgnoringCase()
• assertEqualsWithDelta()
• assertNotEqualsCanonicalizing()
• assertNotEqualsIgnoringCase()
• assertNotEqualsWithDelta()

to replace the respective optional parameters for assertEquals() and assertNotEquals().

[stof]

What about combinations of this flags:

• asserting that 2 arrays contains the same case-insensitive values without a constraint on the order (canonicalize + ignore case)
• asserting that 2 arrays contains the same floats (with delta) without a constraint on the order (canonicalize + delta)

The new methods don't account for all combinations of it (and the comparator library still has them as optional parameters AFAICT, so these combinations would still be supported by the lower-level code, unless the goal is to change these APIs too)

[sebastianbergmann]

I see your point, @stof. However, the combinatorial explosion made possible by mixing the various optional parameters of assertEquals() lead at least to confusion in the past. See #3317, for example. I would really like to reduce the complexity and confusion here. I am also not convinced that there are (enough) real use cases for the combinations to justify them.

I could see, however, keeping the $delta argument (somehow it feels "special") for assertEquals() and assertNotEquals() (removing the need for assertEqualsWithDelta() and assertNotEqualsWithDelta()) as well as adding $delta as an optional parameter to assertEqualsCanonicalizing() and assertNotEqualsCanonicalizing().

[stof]

#3317 is not about assertEquals but about assertContains. AFAICT, ignoreCase should be working fine for arrays (in case they contain strings of course, otherwise there is no point for ignoring the case).

[stof]

note however that I won't fight strongly for keeping the combination of canonicalize and ignoreCase, as I don't remember when I used ignoreCase in my own tests