CODEWITHSUNDEEP
python
El'endia Starman
El'endia Starman

Reputation: 2244

In Python, why does the exact same line of code fail the first time and succeed the second time?

Here's my code exactly as it appears (I'm working on a Numerical Analysis homework problem):

import integration as I
from math import *
import numpy as np

problem = 4

<snip/>

elif problem == 4:

    h = 0.001
    X = np.linspace(0,1, 1//h+2)

    def f(x):
        if x==0 or x==1:
            return 0
        try:
            return x**3/( (1-x)**5 * (e**(x/(1-x))-1) )
        except OverflowError:
            return 0

    try:
        Y = list(map(f,X))
    except:
        print("That failed.")
        Y = list(map(f,X))

    print("Integral: %s"%I.Simpsons_L(Y, h))

Here's the output to shell:

>>> 
That failed.
Integral: 6.49393940227
>>>

If I don't have the except clause in there, then this is the error I get:

Traceback (most recent call last):
  File "<snip/>", line 150, in <module>
    Y = list(map(f,X))
  File "<snip/>", line 145, in f
    return x**3/( (1-x)**5 * (e**(x/(1-x))-1) )
  File "C:\Python33\lib\idlelib\PyShell.py", line 60, in idle_showwarning
    file.write(warnings.formatwarning(message, category, filename,
AttributeError: 'NoneType' object has no attribute 'write'

Edit 1

If I run my program from the command line, this is what I get:

NA_homework3problems.py:145: RuntimeWarning: overflow encountered in double_scalars
  return x**3/( (1-x)**5 * (e**(x/(1-x))-1) )

Clearly, I have a workaround for my problem, so that's not why I'm asking this question. I'm asking why there's a problem in the first place. I am literally executing the same functions twice, and it works the second time but not the first. Also, this issue doesn't come up if I only have map by itself, and if I attempt to use list on the same map twice, the second one is empty, which I think is by design. So, what's going on here?

Upvotes: 1

Views: 862

Answers (2)

El&#39;endia Starman
El&#39;endia Starman

Reputation: 2244

I did some more hunting and pinned down what causes the problem. Somehow, it is not my code, but rather numpy's. The following lines of code are sufficient to throw the error the first time but not the second.

>>> import numpy
>>> 1/numpy.float64(0)
Traceback (most recent call last):
  File "<pyshell#84>", line 1, in <module>
    1/numpy.float64(0)
  File "C:\Python33\lib\idlelib\PyShell.py", line 64, in idle_showwarning
    file.write(warnings.formatwarning(message, category, filename,
AttributeError: 'NoneType' object has no attribute 'write'
>>> 1/numpy.float64(0)
inf

Somehow, sys.__stderr__ gets set to None. I have submitted a bug report for this issue: #5289. Hopefully it'll be fixed soon.

Upvotes: 0

John La Rooy
John La Rooy

Reputation: 304127

Something you've done previously has replaced the file reference in pyshell with None

During your first run through the map, an exception and Warning is being generated

eg

RuntimeWarning: overflow encountered in double_scalars

The exception you are seeing is due to the attempt to print this warning failing.

The normal way warnings work is that once you've been alerted the warning is silenced for the remainder of the session

The second run also triggers the overflow, but since the warning is suppressed it doesn't try to print the error and so the pyshell error is also not encountered

You should look through your code and figure out what you are doing that's blowing away the file variable in pyshell

Upvotes: 3

Related Questions