Improves wait_exponential_jitter

[yxtay]

Summary

Improves wait_exponential_jitter to inherit wait_exponential
- Reuses wait_exponential.__call__ method instead of of duplicating code
- Follows argument names of wait_exponential
- Adds min argument
- Supports supplying max, jitter, min arguments as timedelta

Tests

pytest tests/test_tenacity.py::TestWaitConditions::test_wait_exponential_jitter

Pull request has been modified.

[yxtay]

Is there anything blocking this from being merged? Deprecation concerns? It's been almost 1 year.

[LotfiRafik]

Hi @yxtay ,
in the case min > initial * exp , the result will be wrong (not equivalent to the original formula).
example :
min == 100; max == 10^9; initial * exp == 8; jitter == 10

• Original formula (+adding the min value)

max(max(0, min), min(result, self.max)) == max(max(0, min), min(initial * exp + jitter, max)) == max(max(0, 100), min(18, 10^9)) == max(100, 18) == 100

• Formula in this PR

 max(max(0, 100), min(max(max(0, min), min(initial * exp, self.max)) + jitter, max) == 
 max(max(0, 100), min(max(max(0, 100), min(8, 10^9)) + 10, 10^9) == 
 max(100, min(100 + 10, 10^9) == max(100, 110) == 110

The correct thing to do is to leave the implementation of wait_exponential_jitter separate from the wait_exponential (because its adding jitter before comparing with max, min values..).

[LotfiRafik]

the min part should be equal to max(0, self.min) (in case self.min is negative + follow same formula as wait_exponential)