allow datetime.timedelta for leeway arg

[wbolster]

This fixes #54.

[wbolster]

Btw, the tests directly call into verify_signature(), while this doesn't seem to be part of the public API (not in __all__). Is this intentional?

[wbolster]

Sigh, Python 2.6 doesn't have datetime.timedelta.total_seconds(). I'll add a compatibility hack.

[jpadilla]

@wbolster LGTM. Let's also update the README to document this behavior.

[wbolster]

ok will update the readme with examples

"José Padilla" notifications@github.com schreef op 2 januari 2015 19:59:36 CET:

@wbolster LGTM. Let's also update the README to document this behavior.

---

Reply to this email directly or view it on GitHub:
#56 (comment)

Sent from my Android device with K-9 Mail. Please excuse my brevity.

[mark-adams]

I think this would be cleaner if we kept this line and added timedelta

[mark-adams]

Specifically, it doesn't make sense to add datetime. in 10+ places when we can just add timedelta to this import.

[jpadilla]

Agree with this.

[wbolster]

Okay, will change. I just have a personal distaste for non-module imports. ;)

[wbolster]

I've pushed an updated version. The imports are simpler now, and I've updated the README. The exception approach for Python 2.6 compatibility due to missing timedelta.total_seconds() is left as-is; see the comments on the previous iteration for details.

[mark-adams]

Looks good to me

[jpadilla]

@wbolster thinking about dropping the checks for total_seconds and always using:

leeway = leeway.days * 24 * 60 * 60 + leeway.seconds

[wbolster]

Sure, would be fine with a comment explaining why. But think about those poor microseconds! :)