Coverage decreased (-11.9%) to 65.825% when pulling 91d5355 on murrayrm:fix_conda into 05d6bc5 on python-control:master.

Integration checks look OK. Most are failing because of issue #164 (scipy 1.0 compatibility), but otherwise running properly. There is also an unknown issue with python3.3 that seems to be related to an error while installing package defaults::openssl-1.0.1k-1.

Followup comment: need to rethink this PR a bit. By removing make_version.py, I messed up the automatic versioning that @cwrowley put into place in PR #52. Probably still need to run make_version.py in the conda recipe, but not rely on it for the version information in conda (since __conda_version__ is not supported).

Changes Unknown when pulling 087046f on murrayrm:fix_conda into ** on python-control:master**.

I've updated this PR to leave make_version.py in place, but still use jinja for generating the version information in meta.yaml. I squashed the commits so that the history is clean. This PR will allow conda-build 3.0 to work correctly.

Travis CI checks fail due to SciPy c1.0 incompatibility. This is expected and solved by PR #170.

Changes Unknown when pulling 087046f on murrayrm:fix_conda into ** on python-control:master**.

FWIW, Python 3.3 is end-of-life (see https://devguide.python.org/#status-of-python-branches ). Are you aware of any python-control users requiring it? If not, we might as well drop it.

Travis CI testing for Python 3.3 (and 3.4) has been removed in PR #169. I don't see any need to support Python 3.3 going forward => haven't bother to debug the issue that was described above.