# Empirical Examples

Back in July 2000, I sent these messages to the Python developers’ mailing list. List comprehensions were still a new feature, and I thought the best way to decide on a syntax was to try it out. The idea didn’t catch on, but I still believe very strongly that we ought to field-test language features and additions to libraries before rolling them out, and that such tests are much easier to do than most people believe.

## Message 1

July 11, 2000

I sent the message below to the 11 developers here in the office this morning. They all have 10+ years experience with C, and 2-3 years with C++, Java, or both. None have used Python (I’m working on it :-), but two have extensive Perl experience, and one worked on and with functional languages in grad school. The question was:

``````OK, folks, language question.  Given the statement:

for x in [10, 20, 30]; y in [1, 2]:
print x+y

(note that the second list is shorter than the first),
would you expect to see:

(A) 'x' and 'y' move forward at the same rate:

11
22

(B) 'y' goes through the second list once for each value of 'x':

11
12
21
22
31
32

(C) an error message because the two lists are not the same length?

Thanks,
Greg
``````

Everyone voted (B). As useful as this capability is, I therefore think the proposed syntax is likely to mislead.

## Message 2

July 13, 2000

I gave 20+ grad students in computer engineering and computer science (plus a couple of profs) the questionnaire below. A summary of results is:

• Everyone understood the existing syntax (A).

• Most assumed that that the ‘zip()’ function version would yield cross-product iteration (B). This surprised me a lot, particularly as I think we all agree that ‘zip()’ is the most suggestive possible name for this function.

• Versions (D), (F), and (H) show that semi-colon separated targets imply cross-product iteration, but ‘and’ and ‘while’ imply simultaneous iteration.

• Version (G) shows that using ‘and’ suggests that the sequences being iterated over must have the same length.

The questionnaire and a table of responses are included below (100 characters wide).

Greg

``````The single and double loop below print the output shown:

for x in [10, 20, 30]:          for x in [10, 20, 30]:
print x                         for y in [1, 2, 3]:
print x+y
10                              11
20                              12
30                              13
21
22
23
31
32
33

Match each of the following to its output.
(Note that several examples may print the same thing.)

(1)     (2)     (3)     (4)     (5)     (6)

11      11      11      11      error   ??
22      12      22      12
33      13              21
21              22
22              31
23              32
31
32
33

(A)
for (x, y) in [[10, 1], [20, 2], [30, 3]]:      11      0       0       0       0       0
print x+y

(B)
for [x, y] in zip([10, 20, 30], [1, 2, 3]):     2       6       0       0       0       3
print x+y

(C)
for [x, y] in zip([10, 20, 30], [1, 2]):        0       0       1       4       4       2
print x+y

(D)
for x in [10, 20, 30]; y in [1, 2, 3]:          3       7       0       0       1       0
print x+y

(E)
for x in [10, 20, 30]; y in [1, 2]:             1       0       2       6       2       0
print x+y

(F)
for (x in [10, 20, 30]) and (y in [1, 2, 3]):   7       4       0       0       0       0
print x+y

(G)
for (x in [10, 20, 30]) and (y in [1, 2]):      0       0       1       2       6       2
print x+y

(H)
for x in [10, 20, 30] while y in [1, 2, 3]:     2       7       0       0       1       1
print x+y

(I)
for x in [10, 20, 30] while y in [1, 2]:        0       1       0       7       2       1
print x+y

(J)
for x; y in [10, 20, 30]; [1, 2, 3]:            2       3       0       1       3       2
print x+y

(K)
for x; y in [10, 20, 30]; [1, 2]:               0       1       0       3       5       2
print x+y
``````

Updated: