Wierd bug with flatten and partition

[cdoool]

Hi! First of all, thanks for the great library. Thanks to it I no longer want to migrate back to Scala.

On a first day of using it, I encountered a strange bug:
val = seq([1,2,3,4,5,6]).partition(lambda x:x>2)
(val = [[3, 4, 5, 6], [1, 2]])
But
val = seq([[1,2,3],[4,5,6]]).flatten().partition(lambda x:x>2)
(val = [[3, 4, 5, 6], []])

Surely, it's a bug?

[EntilZha]

Thanks @cdoool for opening an issue! I agree that the behavior seems wrong. I did a little more debugging and I think this example shows that it is due to the sequence being consumed twice in partition. The culprit is that there are two filter calls in the implementation https://github.com/EntilZha/PyFunctional/blob/master/functional/transformations.py#L672. It shouldn't be too hard to rewrite that to do only a single pass/consume the sequence once. I could probably have a fix this weekend and make use of the fancy new travis auto publish to pypi to get the fix out shortly after.

Thanks for the bug report!

In [36]: a, b = seq([[1, 2, 3], [4, 5, 6]]).flatten().partition(lambda x: x > 2)
In [37]: a
Out[37]: [3, 4, 5, 6]
In [38]: b
Out[38]: []
In [39]: a, b = seq([[1, 2, 3], [4, 5, 6]]).flatten().partition(lambda x: x > 2)
In [40]: b
Out[40]: [1, 2]
In [41]: a
Out[41]: [] 

[EntilZha]

@cdoool I believe I've fixed the issue on master. I included the example above in the tests, but it would be great if you can verify for your specific use case that it works that would be great. If everything looks good I can push out a minor release on PyPI

[cdoool]

Can confirm everything works fine now. Many thanks!

[EntilZha]

Thanks, new version should be out on pypi now