Add sample game generators from bimatrix-generators

[oyamad]

For the background, see QuantEcon/GameTheory.jl#53 (comment). Cc: @ptigwe

• Generate doc files

[oyamad]

@mmcky Running python qe_apidoc.py does not generate doc files for bimatrix_generators.py which is located in a subdirectory game_generators in game_theory. Could you give an instruction, to me or to @QBatista, about how to modify qe_apidoc.py?

[coveralls]

Coverage increased (+0.2%) to 95.136% when pulling 0dfa4b7 on oyamad:bimatrix-generators into 49ad385 on QuantEcon:master.

[mmcky]

Thanks @oyamad. @QBatista we can work on this tomorrow, and I will update on how qe_apidoc.py is put together. I haven't documented it well as I was hoping to make it automated (to prevent things not being included) using autodoc.

[mmcky]

The issue is that our qe_apidoc.py only implements one level deep and the directory tree is not recursively searched. To get this PR merged we can add this sub-module into qe_apidoc.py to get the documentation to build.

[oyamad]

If we add something like

game_theory_files += glob("../quantecon/game_theory/[!tests]*/[a-z0-9]*.py")

to

QuantEcon.py/docs/qe_apidoc.py

Line 187 in 931fd11

game_theory_files = glob("../quantecon/game_theory/[a-z0-9]*.py")

would that work?

[mmcky]

That is essentially what needs to be done - except that would list the generators under the game_theory section and not as another sub-directory within game_theory.

@QBatista and I can work on this while he is here. It might be nice to improve the code to make this sort of adjustment easier to do. I am thinking of a simpler json config file that would determine the directory structure and then get python to build the directories and documents from that config file.

[oyamad]

I see. At the moment it is fine with me: the functions in game_theory/game_generators/bimatrix_generators.py are imported into the qe.game_theory submodule (i.e., they can be called by qe.game_theory.*).

[mmcky]

I see. In that case:

game_theory_files += glob("../quantecon/game_theory/game_generators/[a-z0-9]*.py")

should work.

[mmcky]

It's also OK if you want to use your pattern to capture future sub-directories also if you think the sub-modules will always be available at the top level of the game_theory module.

[oyamad]

make html does not work with this approach: ModuleNotFoundError: No module named 'quantecon.game_theory.bimatrix_generators'

[mmcky]

@oyamad this isn't elegant but includes the game_generators subdirectory into the qe_apidoc.py tool. I would like to rewrite this to be recursive and search the directory tree automatically but this manually adds that directory and then adds the bimatrix_generators module to the game_theory toc to keep the toc flat.

As this branch sits in your fork I don't think I can pre-render using rtd (as a version) for you to have a look at but it does render locally (assuming you have your version of quantecon installed or is active in an environment).

[oyamad]

@mmcky Great, thanks so much! It renders perfectly locally, although there is some warning with make html:

:1: (WARNING/2) Inline interpreted text or phrase reference start-string without end-string.

(Does not seem to affect the outcome.)

[mmcky]

Hi @oyamad - Great. I am comfortable with the warning from sphinx for now and happy for this to be merged when you have finalised development.

[oyamad]

There remain some English grammatical issues (common for this PR and QuantEcon/GameTheory.jl#76).

Could you advise us on

• Add sample game generators from bimatrix-generators GameTheory.jl#76 (comment)
• Add sample game generators from bimatrix-generators GameTheory.jl#76 (comment)

Thanks!

[oyamad]

This is ready to be merged.

[oyamad]

Many thanks @QBatista @shizejin @mmcky !