Repository navigation
Python 3.6 compatibility - #717
Conversation
…in 3.6 compatibility
|
@bsipocz - I think I'm getting close to python 3.6 compatibility. I've reached a new error that it looks like you solved in an AstroQuery Issue, but in that Issue you say something I don't understand, "The best and easiest way to deal with it in astroquery to change this test to be pure pytest.". Can you elaborate how I should address this? |
|
@aphearin - #648 got rebased, but is still far behind this one and less consistent with the raw strings. The A third short term workaround would be to disable deprecation warnings that come from pytest to be raised as exceptions. |
…t least the raw string issue is fixed.
…redundant NUMPY_VERSION=prerelease test with python 3.5. This commit should resolve both
|
@bsipocz - I tried to mirror exactly your previous work on the |
| env: NUMPY_VERSION=prerelease SETUP_CMD='test' | ||
| # Python 3.6 currently has a DeprecationWarning related to pytest | ||
| # that we don't understand at the moment. | ||
| - python: 3.6 |
There was a problem hiding this comment.
you need to repeat the matrix entry both in the main matrix part and using the exact setup under allow_failures.
…for the correction @bsipocz
There is only one change I am aware of that will be necessary to gain python 3.6 compatibility: math-mode docstrings need to be changed so that
\\is replaced by\, and all docstrings with such appearances need to be raw strings. I think, but I am not sure, that these changes will address all failures shared by @eteq with me in a recent bug report onv0.5rc1.CC @bsipocz - this PR replaces your #648. Since the
empirical_models.phase_space_modelssub-package received such a massive overhaul with #709, I found it easier to just start over rather than rebase.Currently I am sparing Travis some computation load by having commented out most environments while I'm working on
python 3.6. These will be reinstated prior to merging this PR.