Hash every args including pandas and numpy objects.

[fx-kirin]

I just would like to show you an example of the hash implementation.

#9 (comment)

[shaypal5]

This looks really nice, @fx-kirin !
Once you solve the Python 2.7 test problem (seems like a numpy-pandas version mismatch issue), and as long as you test it thoroughly - I've read the existing two tests, but do try to come up with more - we can merge this! :)

Of course, given that code coverage and quality do not go down. :)

[fx-kirin]

I'm not sure how to fix that dependency problem because I didn't specify any version of them. I think that's the problem of pandas and numpy themselves. The modification of code looks quite ok to me. How do you think?

[shaypal5]

You know what? Screw it. I'm dropping support for Python 2.7. It ended it life cycle and the community should move away from it anyway.

[shaypal5]

Merged! I think it will pass all the tests and is A-OK test coverage and code quality wise!

Thank you for the contribution!

[shaypal5]

Hey @fx-kirin ,

Unfortunately I had to roll back all hash-related additions due to issues #18 and #19 .
Any future attempt to implement this kind of feature should explicitly treat those issues, including testing for them using the code supplied by the reporter of the issues.

The hash branch contains all the changes before the revert commit.

Bummer. :/
But maybe this can be solved! :)