Bug in datetime lib causes incorrect timedelta value

[oldwizard1010]

Bug report

Bug description:

Problem with datetime lib

# Add a code block here, if required
import time
from datetime import datetime

x1 = (datetime.fromisoformat('2025-01-08T00:00:01') - datetime.utcnow()).seconds
time.sleep(5)
x2 = (datetime.fromisoformat('2025-01-08T00:00:01') - datetime.utcnow()).seconds
print(f"x1 should be less than x2, but x1={x1}, x2={x2}; x1<x2={x1<x2}")

CPython versions tested on:

3.12

Operating systems tested on:

Linux

[picnixz]

For me, it works as expected. Precision is a bit lost but it's still correct (the delta is 5 seconds, so it's a <= and not an exact <):

If you want to check < you should also consider the microseconds part.

[vadmium]

What was the “incorrect” value? Maybe this was done as seconds wrapped between zero and a whole day? Try using total_seconds.

[ericvsmith]

Closing since there's no follow-up.