Test on pypy2/3 plus irrelevant bug on tests discovered by the build

[Digenis]

Tests succeed on pypy2/3 (thus added to travis)
except for some tests on sets
that were accidentally written as deterministic.
(I fixed them in the same PR to keep the build clean)

---

By the way,
while set(seq('abc')) gives the expected in the shell,
in tests I get an empty set and can only use .the set() method.
Probably assertSetEqual(result, set(expected)) doesn't work for the same reason.

Looks to me like a bug you may want to investigate.

[coveralls]

Coverage remained the same at 98.41% when pulling e6fd017 on Digenis:master into 43fe6a6 on EntilZha:master.

[EntilZha]

Good catch on the tests and thanks for adding PyPy.

My intuition is that what behavior you are seeing is related to how I did the initial implementation of generators from #17 on master. The problem at the moment is that an iterator is returned on transformations, so if that result is iterated on more than once, it returns nothing the second time. What I am unsure of is that the set result doesn't look like its being iterated over more than once... In any case this overall issue is being worked on in #20 and the lineage-rewrite branch.

PR looks good to go though.

[EntilZha]

Would you mind making an issue with a code example of what is breaking? I am trying to reproduce with no luck.

[Digenis]

Try this, it results in an empty set.
Dropped to pdb and retried without being able to reproduce it.

diff --git a/test/test_functional.py b/test/test_functional.py
index 78a0342..149dfe2 100644
--- a/test/test_functional.py
+++ b/test/test_functional.py
@@ -222,7 +222,7 @@ class TestChain(unittest.TestCase):
         result = seq([1, 1, 2, 3, 3]).union([1, 4, 5])
         expect = [1, 2, 3, 4, 5]
         self.assert_type(result)
-        self.assertSetEqual(result.set(), set(expect))
+        self.assertSetEqual(set(result), set(expect))

======================================================================
FAIL: test_union (test.test_functional.TestChain)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/digenis/src/ScalaFunctional/test/test_functional.py", line 225, in test_union
    self.assertSetEqual(set(result), set(expect))
AssertionError: Items in the second set but not the first:
1
2
3
4
5

Name               Stmts   Miss  Cover   Missing
------------------------------------------------
functional             8      0   100%   
functional.chain     307     12    96%   8-14, 121, 129, 870, 1004, 1012
------------------------------------------------
TOTAL                315     12    96%   
----------------------------------------------------------------------
Ran 79 tests in 0.712s

FAILED (failures=1)

[Digenis]

Also, I noticed that the same pattern for dicts (not using assertDictEqual) so I pushed once more.
I can't reopen the PR, maybe you can try pulling like this to get the last commit:
git pull git@github.com:Digenis/ScalaFunctional.git master

[EntilZha]

I run the same thing without errors, might running on pypy vs standard python matter? I can either pull that commit or can merge another PR if you would like, either is fine with me.

[EntilZha]

Ya, I just installed pypy and it fails with that, but not with regular python. I can take a look and dive deeper later today/tonight.

[Digenis]

No need for PR formalities for a follow up committed so soon :).
I tested on pypy3. I didn't try it in regular python. That surprises me.

[EntilZha]

Made an issue after some extensive debugging at #25