Better way to structure the repo?

[bakerwho]

Perhaps it's just me, but it looks pretty hard to navigate through the code while reading through the book. I have to open and read agents.py to see that it discusses code from Chapters 1 and 2.

There's a notebook called intro.pynb that gives some information about the reasons for this -

A typical module in the repository has the code for a single chapter in the book, but some modules combine several chapters.

But this notebook is not referenced in the README. Besides, it may be nice to have a companion to the book that links, chapter by chapter, section by section, what files or notebooks might be relevant.

Similarly, the instructions to deal with the aima-data submodule are in SUBMODULE.md. But this is not mentioned anywhere in the README. The emptiness of the aima-data subfolder after a git clone might be confusing for newcomers to git. If they move files into the folder from local downloads, it might also complicate the merging of their changes (not 100% sure about this, but it looks like it might be a problem).

I'd like to open a discussion on this before I make any changes, but what do you think, @norvig, @MrDupin?

[antmarakis]

You are right that we do not have a clear installation guide. I have amended that in #773, thanks for pointing it out!

The instructions in SUBMODULE are simply for updating the submodule to the latest version, so that users can download the correct datasets. It is not something that users need to do themselves, it is more of a reminder for admins.

About the structure, I think it's fine as it is. We already have the README that points people to the right direction. It is up to Mr. Norvig, but I personally wouldn't change this.

[JaeDukSeo]

I tried to clean it up, but it wasn't the right approach. So I guess we'll have to wait.

[bakerwho]

@MrDupin, @norvig Might it make sense to put the notebooks in a folder called 'notebooks' and the .py files in a folder called 'bin' or 'pycode' or 'scripts'? The repo is still a bit messy. Scrolling through it, one finds files like pytest.ini and requirements.txt in midst of ipynb files that explain concepts! Is there a reason such restructuring has not already been done?

If we're not taking any further steps in this direction, I reckon this issue can be closed.