expect([
row[‘blah’][0][1],
row[‘blah’][1][1],
row[’blah’][2][1],
row[‘blah’][3][1],
row[‘blah’][4][1],
]).to contain_exactly(
a.name(user), # “John doe”
c.name(user), # “John doe”
e.name(user), # “John doe”
b.name(user), # “John doe”
d.name(user), # “John doe”
)

(Note: The comments are mine.)

See the problem? No, not the ugly code (which is actually worse than what i posted here).

It’s using the same ridiculous getter (if you can call it that) that pulls a name out of the passed user object, and then expecting each row to have that name, in order. Not that order matters when they’re all the same.

Upon inspection, all objects created by the spec have the exact same name, so the above test passes (as long as there are 5 rows). It passes, but totally not because it should: those aren’t the objects that are actually in the table. All of the specs — all 22 of them — only check for that shared name on various rows, and no other data. And it’s not like this is the only issue, either.

Fuck me these are bad.

And this guy is a senior dev earning significantly more than me. Jesus what the fuck Christ.