This commit also:
- Dramatically increases the number of unit tests , mostly by borrowing
from the standard library's unit tests for math.isclose().
- Refactors approx() into two classes, one of which handles comparing
individual numbers (ApproxNonIterable) and another which uses the
first to compare individual numbers or sequences of numbers.
When defining a fixture in the same module as where it is used, the
function argument shadows the fixture name, which a) annoys pylint and
b) can lead to bugs where you forget to request a fixture into a test
method.
This allows one to define fixtures with a different name than the name
of the function, bypassing that problem.
This was a challenge because it had to work in python2 and python3,
which have almost opposite unicode models, and I couldn't use the six
library. I'm also not sure the solution I found would work in python3
before python3.3, because I use the u'' string prefix which I think was
initially not part of python3.
When we have a metaclass which returns something truthy (like a method) in its
__getattr__, we collected the class because pytest thought its __test__
attribute was set to True.
We can work around this to some degree by assuming __test__ will always be set
to an explicit True if that's what the user has intended, and if it's something
other than that, this is probably a mistake.
Fixes#1204.
When an object has a custom __getattr__ which always returns a non-int, we
tried to get compat_co_firstlineno from it and checked it was a integer, which
caused an exception if such a class is mistakenly collected.
If we still mistakenly collect such a class (which is likely to be something
other than a test), we now skip it with a warning (because it probably has an
__init__) instead of producing an error.
See #1204.
Fixes issue 331
previously to this change the collection code would issue a warning for
when ever it encountered a variable that looked like a test but wasn't a
function saying that it wouldn't collect it because it wasn't a function.
This fixes the logic so that if that warning is issued it really isn't
collected.
However previously special cases existed to support tests that were
created using functools.wraps and functools.partial. So the condition for
issuing that warning has been updated to take that in to account
Also try the old way of detecting functions just for proper integration
with mock.path in python 2.7 the get_real_func returned the unbound method