Author ncoghlan
Recipients Amber.Yust, Andreas.Pelme, Hanxue.Lee, Lakin.Wecker, alex, belopolsky, cvrebert, dstufft, ethan.furman, georg.brandl, gwrtheyrn, lemburg, merwok, ncoghlan, pitrou, r.david.murray, shai, skip.montanaro, tim.peters, yselivanov
Date 2014-03-07.14:22:46
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <CADiSq7c=Uom3iNe1weAZJa+0FS6RQ+DQMV=j-H9pA3c7bWy+Lg@mail.gmail.com>
In-reply-to <1394198362.0.0.893392430588.issue13936@psf.upfronthosting.co.za>
Content
Structured data is just a shorthand way of referring to any Python object
which is neither a number or a container and exhibits the default boolean
behaviour where all instances are true.

The problem datetime.time is both that its current behaviour is internally
incoherent (whether or not an aware time is false depends on the current
timezone in unpredictable ways) and *also* inconsistent with its other
behaviours that indicate it should be handled as a non-numeric value. Since
it isn't a container either, standard conventions suggest that it should
always be true. No *compelling* justifications for its atypical behaviour
have been presented, just a case of Tim wanting to leave the door open to
adding modular arithmetic directly on time instances.

I suggest it makes far more sense to instead eliminate the quirky behaviour
entirely and instead provide an easy way to convert a time to a timedelta
relative to midnight.
History
Date User Action Args
2014-03-07 14:22:46ncoghlansetrecipients: + ncoghlan, lemburg, tim.peters, skip.montanaro, georg.brandl, belopolsky, pitrou, merwok, alex, r.david.murray, cvrebert, ethan.furman, gwrtheyrn, Lakin.Wecker, yselivanov, shai, dstufft, Andreas.Pelme, Amber.Yust, Hanxue.Lee
2014-03-07 14:22:46ncoghlanlinkissue13936 messages
2014-03-07 14:22:46ncoghlancreate