Dave Rolsky schreef:
> I think the fix is to pick one way that works "good enough" as the
> default, and document the whole thing really well so that people who need
> a different answer know how to get it.
As long as h/m/s in utc and y/m/d in local is the default option, I'm
not really unhappy.
In the cvs version of Changes, I see:
- Also partially revert the change in 0.28 for subtract_datetime().
To be consistent, the subtraction uses the local portions for year,
month, & day, and UTC for hour, minute, and second.
Which sounds good.
Eugene
