Inputs - JDFTx IO Module

[benrich37]

Summary

Major changes:

• feature 1: inputs.py module containing JDFTXInfile class to give pythonic representation to the inputs for a JDFTx calculation
• feature 2: "helper" module generic_tags.py holding "Tag" objects (AbstractTag and inheritors of AbstractTag) to represent the different data structures JDFTx expects for inputs
• feature 3: "helper" module jdftxinfile_master_format.py to create proper "Tag" objects for input tags
• feature 4: "helper" module jdftxinfile_ref_options.py to hold lists of acceptable input strings for different input tags (ie list of acceptable XC functionals for the tag "elec-ex-corr")

[computron]

@mkhorton , are you OK with merging this change?

@shyuep I am not sure why some of the optional checks are not running, I started the tests a couple of days ago.

Looks like all changes are restricted to the io/jdftx module, so not much likelihood of things breaking

[mkhorton]

Thanks for your patience with this one, long PRs adding a big new feature are always more difficult to review. I think this nicely self-contained and is in good shape to merge in. We can save further improvements for follow-up PRs that we can review/merge more quickly, and we can take the same approach with #4190 too.

One thing that wasn't obvious to me, why the /tmp directory in the test files? I couldn't see where it was used in the diff.

[benrich37]

Thanks for your patience with this one, long PRs adding a big new feature are always more difficult to review. I think this nicely self-contained and is in good shape to merge in. We can save further improvements for follow-up PRs that we can review/merge more quickly, and we can take the same approach with #4190 too.

One thing that wasn't obvious to me, why the /tmp directory in the test files? I couldn't see where it was used in the diff.

Awesome thank you! Sorry for the new push - had a minor bug from inconsistent tracking of units. And the tmp directory is used to hold files that are created and deleted during the tests (ie testing that an input file can be written and then read correctly). I thought it would make the tests run quicker to have the directory exist permanently instead of created when needed but I can change that if the alternative is better.

Edit: The new push is actually restricted to #4078 -

We can save further improvements for follow-up PRs that we can review/merge more quickly

So edits for minor bugs I should keep within #4078 and leave #4189 and #4190 untouched, right?

[mkhorton]

I thought it would make the tests run quicker to have the directory exist permanently instead of created when needed but I can change that if the alternative is better.

It'd be better to remove it :) Creating a directory doesn't take much time, and might add some confusion to others to see it there. Tests should do "set up" and "tear down" as needed.

So edits for minor bugs I should keep within #4078 and leave #4189 and #4190 untouched, right?

As long as you merge in changes from the main branch (or rebase) #4078 once #4189 and #4190 are merged, or you can create new PRs. Either way is fine, I will try and be responsive!