Can you please include the code in-line? What exactly do you mean by 'break down'?

I meant to add pictures, here they come.

By "break-down" I mean the graph doesn't plot past 10^-1 on the y-axis

With what version? The code you provided works fine with 1.4.3 and 1.5.0. (And on a side note, you shouldn't really use pylab.)

It looks like the version on Windows is 1.4.3 and Xubuntu's is 1.3.1. I downloaded the 1.5.1 and re ran the code and the graph is the same. Also instead of getting 1.5.1 when I execute this, python -c 'import matplotlib; print(matplotlib.version)' , I get 1.3.1 so for some reason the version hasn't changed.

@QuLogic Why do you recommend against the use of pylab?

pylab shouldn't be used for scripting. It really should only be for
interactive situations where you want conciseness. We have been working on
getting rid of scripting examples of pylab, because "explicit is better
than implicit" and "namespaces are a honking good idea". The other fun
thing that happens is that np.min(), np.max(), np.any() and np.all()
overrides the built-in min(), max(), any() and all(), which can cause very
subtle issues in some edge cases.

On Wed, Jan 20, 2016 at 6:20 PM, kxf010 [email protected] wrote:

@QuLogic https://github.com/QuLogic Why do you recommend against the
use of pylab?

—
Reply to this email directly or view it on GitHub
#5885 (comment)
.

They are not even that edge:

In [26]: min((i for i in range(3)))
Out[26]: 0

In [27]: import numpy as np

In [28]: np.min((i for i in range(3)))
Out[28]: <generator object <genexpr> at 0x7fcd58ccfca8>

or more fun

In [32]: bool(np.all((0 for j in range(5))))
Out[32]: True

We really should get around to taking advantage of the single-dispatch
system to register those functions for array-like inputs and get pylab to
not clobber the builtins.

On Wed, Jan 20, 2016 at 8:01 PM, Thomas A Caswell [email protected]
wrote:

They are not even that edge:

In [26]: min((i for i in range(3)))
Out[26]: 0

In [27]: import numpy as np

In [28]: np.min((i for i in range(3)))
Out[28]: <generator object at 0x7fcd58ccfca8>

or more fun

In [32]: bool(np.all((0 for j in range(5))))
Out[32]: True

—
Reply to this email directly or view it on GitHub
#5885 (comment)
.

Thanks! Also, are there any known workarounds for when an older version is in use?

yeah, don't use pylab... (actually, there is a real workaround, but it is
extremely convoluted. We implemented it in matplotlib for certain internal
uses, but perhaps it makes sense to export it somehow?

On Thu, Jan 21, 2016 at 9:23 AM, kxf010 [email protected] wrote:

Thanks! Also, are there any known workarounds for when an older version is
in use?

—
Reply to this email directly or view it on GitHub
#5885 (comment)
.

See https://stackoverflow.com/questions/18774388/re-import-aliased-shadowed-python-built-in-methods for how to get the original functions back.

However, this has been side tracked from the original bug. What version of mpl are you using?

I'm using both 1.5.1 and 1.3.1. My goal is to try and get a working bit of code on 1.3.1 so users will be able to use the code without having to update matplotlib as 1.3.1 is the latest in Ubuntu's Software Center(USC). For some reason it hasn't actually updated to 1.5.1 on the USC.

It looks like there is not a direct work-around with 1.3.1. The best course of action would be to do your histograming via numpy

bins, edges = np.histogram(data)

and then ship a version of the code in #643 (comment) to do the plotting.

We definitely will not be releasing a 1.3.2.