Figure out what to do about multiple invocations of the same function on one line

[Norbo11]

def sleep(i):
    time.sleep(i)
    return i

def f():
    x = [sleep(1), sleep(1), sleep(1)]

PyFlame reports one unique stack trace here:

f:1 (3000 ms) -> sleep:1 (3000 ms)

On the other hand:

def sleep(i):
    time.sleep(i)
    return i

def f():
    sleep(1)
    sleep(1)
    sleep(1)

PyFlame reports three unique stack traces:

f:1 (1000 ms) -> sleep:1 (1000ms)
f:2 (1000 ms) -> sleep:1 (1000ms)
f:3 (1000 ms) -> sleep:1 (1000ms)

The second case is correct as the first line of sleep executes for an average of 1 second.

The first case is incorrect as the first line of sleep is assumed to execute for an average of 3 seconds.

Ideas:

Use static analysis to somehow detect these cases and distinguish between them
Figure out how pyspy would behave, once intermediate output is available. Maybe it solves this issue for us.