ENH: adding __str__ and __repr__ methods to classes

[sglyon]

This is not quite finished.

I am just opening this pull request so I can keep track of notes I collect as I go.

I will leave a comment saying it is ready to go at the appropriate time.

[sglyon]

Comments so far:

• I don't have a great __str__ method for ARMA. Any suggestions are welcome
• I am still struggling with how to represent attributes of the class that are functions. I would really like the __repr__ methods to return a copy/pasteable string that would re-create the object, but I don't know if this is possible. I tried using inspect.getsource, but that didn't work so well because it just returns the entire contents of each line containing the definition. For example, when I asked for the contents of ConsumerProblem.du it gave me this whole line, which also contains the definition of ConsumerProblem.u. I could manually parse it, but that seems like more work than we should be doing for this convenience feature and very fragile.

[coveralls]

Coverage decreased (-0.85%) to 92.37% when pulling 505033f on repr into 26bc87e on master.

[sglyon]

More comments:

• Because their arguments are mostly arrays, the idea of having __repr__ return a string that can recreate the object is not feasible for the Kalman, LAE, LQ, LSS, RBLQ, and ECDF. I have just implemented somewhat meaningful __str__ methods and am having __repr__ just call __str__ for now.

[sglyon]

@oyamad

The DiGraph class was contributed by you (thanks!) and I have I implemented very basic versions of these methods, feel free to change them if you think a different message would be better

[sglyon]

@davidrpugh

Given that you spent so much time on this code, I wanted to give you a chance to decide what should go into these method for the classes you wrote. Thus, I have left the following classes without these methods for now:

• Model (in model.py)
• CESModel (in ces.py)
• CobbDouglassModel (in cobb_douglass.py)
• ImpulseResponse (in impulse_response.py)
• IVP (in ivp.py)

Please let me know if you would like to write these methods yourself if you want me to try to figure out something reasonable.

Thanks!

[sglyon]

At this point we just need to implement the display methods for the classes indicated in the checklist in the previous comment to @davidrpugh . All other classes are ready to go.

So, once we have finished with those, this PR should be ready to merge.

[coveralls]

Coverage decreased (-1.67%) to 91.54% when pulling 6e5e83d on repr into 26bc87e on master.

[davidrpugh]

@spencerlyon2

Thanks for the heads up. I will look at this later this evening.

[davidrpugh]

@spencerlyon2

I have committed the changes locally and pushed them to my fork of the quantecon repo, but the commits are not showing up here. Somehow the repr branch in my fork seems to have become detached from the repr branch here. Thoughts?

[sglyon]

Thanks for doing that! That's great

The little note below your comment says Add more commits by pushing to the repr branch on QuantEcon/QuantEcon.py.

I'll just pull changes down from your repo and push to this one.

[davidrpugh]

@spencerlyon2

Weird. I didn't see any such note. In any case I initially tried

git push -u https://github.com/QuantEcon/QuantEcon.py.git/repr

this failed due to insufficient permissions.

[coveralls]

Coverage decreased (-2.02%) to 91.2% when pulling 57f1b2e on repr into 26bc87e on master.

[davidrpugh]

Also I haven't yet added anything to the IVP class yet. The constructor takes functions as inputs and there aren't any parameters per se. Do you have any ideas?

[sglyon]

I'm not sure you have push permissions to QuantEcon/QuantEcon.py, that might why that failed. It was the right idea though.

If there really isn't any useful information to print for the ivp class you could leave it without a display method -- I'd be totally fine with that.

[oyamad]

@spencerlyon2 Re: DiGraph

Very good, thanks!

[sglyon]

@jstac I think this is ready.

Feel free to merge when you have had a chance to look over it.

[jstac]

This is really neat, thank you!