[Python] Tuple comprehension creates generator; List comprehension evaluates all elements...

I am using a long ad hoc script for exploratory data analysis -- not for tool development. The script has gotten quite long, so I've taken to deling ephemeral variables to keep the Spyder Variable Explorer clean. I've done this all over the script.

I tried to be streamline the script by coding some loops as tuple comprehensions, thus eliminating the need for an extra line of code to del the iteration variable. Here is an example of three ways to iterate through figures to clear the plots:

# Generate the figures
import matplotlib as mpl
import matplotlib.pyplot as plt
plt.close('all')
plt.scatter([1,2,3],[4,5,6])
plt.figure()
plt.scatter([1,2,3],[6,5,4])

# 3 ways to clear the figures

if True: # Use Tuple Comprehension
( plt.figure(iFig).clf() for iFig in plt.get_fignums() )

elif True: # Use List Comprehension

[ plt.figure(iFig).clf() for iFig in plt.get_fignums() ]

else: # Don't use comprehension

for iFig in plt.get_fignums(): plt.figure(iFig).clf()
del iFig # The extra line of source code I want to avoid


The 3rd and final option is the one I've been using. The 1st and 2nd options are my attempts at tuple comprehension and list comprehension.

The tuple comprehension returns a generator object and doesn't actually execute evaluate the invocation to clf unless I assign the tuple to x and execute next(x) twice. I can see the figures clearing each time.

This is unnecessary for the list comprehension.

I started off with tuple comprehension with the rationale that I can avoid using a mutable container if I'm not going to use the contents anyway.

Why does the tuple comprehension yield a generator that must be iterated through while list comprehension evaluates all the elements needed to build the list right away?

Is there a known Python rule about list and tuple comprehension that would have allowed me to foresee this?

Continue reading...