Defining Roadmap to Slycot 0.7.0

[KybernetikJo]

I would like to define a Roadmap to the next version of Slycot. Because with #201 and #203 there will be new routines, I guess the version should be increased to v0.6.0!

First a merge of

• Bugfix ab13bd #200
• Add ab04md #201
• Add sb10yd #203

would be nice. I can revise #201 and #203 again, but basically the wrappers work.

I guess #200 needs a final decision. Furthermore, some other Slycot routines identified by @roryyorke in #200 may need a fix.

---

First, I would like to ask if a sphinx documentation is actually desired?

If the answer is yes, I would like to merge a sphinx documentation:

• [ ] Add sphinx doc #202. This Draft/PR works and would be a start.

After that, the following could be done:

• update the counts of the wrapper Slycot/slycot/__init__.py, the numbers are out of date (see images attached). Refactor __init__.py #205
• improve and revise Slycot/slycot/__init__.py. Refactor __init__.py #205
• [ ] get rid of Slycot/slycot/example.py, add Slycot/examples folder as in python-control and move all examples over there. (the folder is already in Add sphinx doc #202)
• refactor and improve docstrings to meet numpydoc
  • bring analysis.py closer to numpydoc, Docstring standards in Slycot #100, Change analysis.py to numpydoc style #206 (this file is very different from the numpydoc specification)
  • bring math.py closer to numpydoc, Docstring standards in Slycot #100, Change math.py to numpydoc style #211
  • bring synthesis.py closer to numpydoc, Docstring standards in Slycot #100, Change synthesis.py to numpydoc style #212 (LGTM)
  • bring transform.py closer to numpydoc, Docstring standards in Slycot #100, Change transform.py to numpydoc style #213
• refactor and improve tests to get a common structure. Define a common structure?
  • get rid of unittest.TestCase, Add sb10yd #203 (comment), Add ab04md #201 (comment) Replace unittest calls with pytest #207
  • [ ] define a common test structure
  • [ ] study pytest structure and usage in python-control control pytest
• provide a contribution guide
  • [ ] we could add a full example of how to create a wrapper. However, it is quite a bit of work.
  • [ ] a slycot template for a .pyf would be nice.
  • [ ] a slycot template for a numpydoc would be nice.
• [ ] finally, discuss and improve the slycot sphinx documentation in general.

Remark:
Things that are marked as ~tasks~~ will most likely be postponed to Slycot v0.7!

What do you think?

[roryyorke]

Thanks for all the work @KybernetikJo. Thanks also @bnavigator for reviewing and merging---I don't have much free time at the moment.

FWIW, you might be interested in python-control/python-control#928 , though I don't know if it's gone anywhere.

[KybernetikJo]

@bnavigator @roryyorke

Because I have less time than expected, I would like to move the addition of the sphinx documentation to a milestone 0.7.0! That includes the movement of examples and also the pytest file structure refactoring.
Otherwise, the PR #202 could become a showstopper for a Slycot 0.6.0 release.

In addition, I would like to postpone #208 to milestone 0.7.0
Numpy 2.0.0 is going to have a lot of updates to f2py. It could be a good time to have a fresh look at it.

Defining Roadmap to Slycot 0.7.0:

This would be my wish list, but I have to add "Resolve or postpone these issues before releasing 0.7.0."

• Add sphinx documentation Add sphinx doc #202
• Further improvements of docstrings, numpydoc Docstring standards in Slycot #100
• Movement of examples
• Discussing and refactoring of the pytest file structure
• Call by reference thing SLICOT routines alter input parameters #208 (wating for numpy 2.0.0)
• New Wrappers, Subspace Identification (IB) routines #169 (I would like to add SID methods)
• Wheels and build system, Build wheels in CI and publish them #164 (Not sure how helpful I can be. The meson transition of numpy, scipy, pandas and matplotlib was successful, fun read https://labs.quansight.org/blog/building-scipy-with-flang)

[bnavigator]

Sounds good to me! I'd rather get the existing improvements into a 0.6 and get work on a 0.7 started, than waiting for perfection.

What do we really want to have in 0.6 already? Maybe at least a hint, that as of now any wrapper "may or may not" alter input parameters.

[KybernetikJo]

What do we really want to have in 0.6 already?

I think all PRs since v0.5.4! Three new wrappers and four existing wrappers have been added to Slycot/__init__.py.

Maybe at least a hint, that as of now any wrapper "may or may not" alter input parameters.

Where do you want to place this hint? Readme.md? Release Notes? Would that be enough?

---

Current state of Slycot.

[bnavigator]

Where do you want to place this hint? Readme.md? Release Notes? Would that be enough?

Yes, both is a good idea and would be enough in my eyes.