Create a BoxPlot class about axelrod

[drvinceknight]

Tagging myself in as I think I can help with this one.

[meatballs]

I've made a start on this one - branch 88 in my fork.

[drvinceknight]

👍 Will try to get to it.

[drvinceknight]

Also, pretty sure we can add matplotlib as a dependency on travis... Let me go check that...

[drvinceknight]

Yeah looks fine:

http://docs.travis-ci.com/user/languages/python/

Here's a SO question (mainly due to a problem with python 3 so not relevant): http://stackoverflow.com/questions/1739494/travis-ci-matplotlib-dependency-and-python3

[meatballs]

I'm hitting the matlibplot problem again - must be something I don't understand about the library.

Line 47 in run_tournament.py ought to give the same results as line 48, but doesn't.

[drvinceknight]

I've got some time, so taking a look now.

[drvinceknight]

Found the error somewhere, I think your numpy array is carrying out integer division... Finding where that happens now.

[drvinceknight]

Think I've fixed it: sending you a PR (apologies in advance if I just confused myself).

[drvinceknight]

Ok, I think I have to actually 'request that you pull', I can't seem to get github to send you a nice pull request as I did not fork your fork...

https://github.com/drvinceknight/Axelrod/tree/88

Let me know if that doesn't work :) 👍

[drvinceknight]

That branch failed on travis for me by the way (but I just checked that outputs where the same, didn't run tests).

[meatballs]

I'll have a look some time tomorrow - thanks for the help

[drvinceknight]

My pleasure (if I understood the issue correctly obviously).

On Sat, Mar 7, 2015 at 9:38 PM Owen Campbell [email protected]
wrote:

I'll have a look some time tomorrow - thanks for the help

—
Reply to this email directly or view it on GitHub
https://github.com/drvinceknight/Axelrod/issues/88#issuecomment-77711428
.

[meatballs]

Yep - that works!! (But why did that need adding to the method, when it's not required within run_tournament.py)?

I'll finish this off and submit a PR when I'm done

[drvinceknight]

Yep - that works!!

Awesome!

(But why did that need adding to the method, when it's not required within run_tournament.py)?

Because at the top of run_tournament we tweak the division: from __future__ import division. I wouldn't want to do this throughout axelrod but in a script that makes use of the library I thought it was fine. When we eventually package this so that people can run all sorts of different types of tournaments I'd want them to know that 'normal' Python 2 division is still going (even though it's a pain).

Any chance of you de-numpying? No problem if not, can get to that at another time.

[meatballs]

Agggghhhhhhh!!!! Missed that subtlety entirely!!!

OK - I'll sort out the de-numpification and the try-catch for matplotlib before I create the PR.

[drvinceknight]

(By the way: the test suite you wrote, really is excellent I'm just writing a bunch of tests for #52 and it makes it really easy: thanks!)

Agggghhhhhhh!!!! Missed that subtlety entirely!!!

Hehe, yeah pretty much my first go to when debugging Python: 'integer division'?

Really no problem if the de-numpy thing doesn't happen, I expect that'll be a weekend job to go through everything at some point...

[meatballs]

On second thoughts, I'll leave the de-numpification for a separate PR.

[drvinceknight]

Yeah, completely fine: we're in transition :D

On Sun, Mar 8, 2015 at 2:55 PM Owen Campbell [email protected]
wrote:

On second thoughts, I'll leave the de-numpification for a separate PR.

—
Reply to this email directly or view it on GitHub
https://github.com/drvinceknight/Axelrod/issues/88#issuecomment-77753430
.

[meatballs]

(By the way: the test suite you wrote, really is excellent I'm just writing a bunch of tests for #52 and it makes it really easy: thanks!)

You're welcome. I had to write it for myself in order to understand properly what most of the code was actually doing!!