A generator expression produces values from an iteration specification without first building the complete result collection. It resembles a comprehension, but uses parentheses rather than the square brackets of a list comprehension. The expression returns a generator object, not a list; a `for` loop can consume its values. section 8.9.1 section 8.4
x = [1, 2, 3, 4]
x_squared = (item * item for item in x)
x_squared
<generator object <genexpr> at 0x102176708>
for square in x_squared:
print(square,)
1 4 9 16
Squaring four items produces a generator object. Iterating over that object prints `1 4 9 16`; merely displaying the object does not display those values. section 8.9.1
Count without building a list of counts
Given a list of lines, `sum(len(line.split()) for line in lines)` adds the word count for each line. The analogous `sum(len(line) for line in lines)` counts characters. The generator expressions avoid constructing separate lists of per-line counts, but `readlines()` has already loaded the lines into memory. section 8.13.2
Check dictionary entries
`any(not isinstance(_, str) for _ in self.keys())` tests whether any key is not a string; the same pattern can check values. Here `_` is a conventional name for a throw-away iteration variable. section 17.11.2
Choose a generator expression when values can be consumed as they are produced, especially when retaining a complete result list would waste memory. It does not make the rest of a program memory-efficient: reading an entire file or creating another large list can still dominate memory use. section 8.9.1 section 11.5.2
