Averages don't take into account milliseconds

[ericfrederich]

Averages seem to just copy the millisecond portion instead of averaging them.

For two instances of Pendulum p1 and p2 I believe it should hold that p1.average(p2) == p2.average(p1)

This is not currently the case. In fact, averaging two times that are only 1 second apart will yield one of the source times rather than a midpoint.

import pendulum
import time

p1 = pendulum.now()
time.sleep(1)
p2 = pendulum.now()

a1 = p1.average(p2)
a2 = p2.average(p1)

print(p1)
print(a1)
print(a2)
print(p2)
print(a1 == a2)
print(p1 == a1)
print(p2 == a2)

example output:

2017-08-23T12:01:32.809279-04:00
2017-08-23T12:01:32.809279-04:00
2017-08-23T12:01:33.821634-04:00
2017-08-23T12:01:33.821634-04:00
False
True
True

[sdispater]

Yes, you are definitely right.

I just checked the code and average() only averages at the seconds level and not microseconds.

It's an easy fix so it should not be too long before it makes it to the next minor release

[ericfrederich]

@sdispater, great! Add a test too if you can. I currently have datetime code that recursively bisects intervals. I'd like to switch to pendulum but obviously cannot until this is fixed.

[ericfrederich]

Pinging @sdispater ... I wrote a fix and some tests.

Things in 2.0 branch are refactored so it's not a clean merge. I can merge conflicts and get 2.0 branch up to date once this is merged into master.

[ericfrederich]

Look good @sdispater ?

[sdispater]

@ericfrederich Yes, your PR fixed this. However I will close this once I publish the next release.