Open Stack

On Mon, 2014-11-24 at 13:19 -0500, Jay Pipes wrote:
> I think pointing out that the default failure 
> message for testtools.TestCase.assertEqual() uses the terms
> "reference" 
> (expected) and "actual" is a reason why reviewers *should* ask patch 
> submitters to use (expected, actual) ordering.

Is there any reason for this specific ordering? Not sure about others,
but I tend to write equality comparisons like this

    if var == 1:

instead of

    if 1 == var:

(although I've seen the latter in C code before).

This gives rise to

    assert var == 1

or, moving into `unittest` domain

    assertEqual(var, 1)

reading it as 'Assert `var` equals 1', which makes me wonder why the
`assertEqual` API is defined the other way around (unlike how I'd write
any other equality check).

Nicolas