Author lemburg
Recipients belopolsky, bozo.kopic, ghaering, lemburg, serhiy.storchaka, xiang.zhang
Date 2017-01-11.17:16:25
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <480c5900-089f-847e-25a5-3dc52668b78c@egenix.com>
In-reply-to <1484150669.73.0.569277648635.issue29099@psf.upfronthosting.co.za>
Content
On 11.01.2017 17:04, Xiang Zhang wrote:
> I am not sure it's worth to make it even optional in 3.7. Discarding tzinfo sounds weird, and worse, the behaviour is not deterministic, it could also fail with an exception.

Best practice is to store all date/time values using UTC (or some
other fixed timezone) in databases and to manage the timezone/locale
information elsewhere.

The TZ info then becomes redundant and only results in more
database space being used as well as slower queries (due to the extra
TZ calculations being done; many databases internally store the
values as UTC to avoid having to do TZ calculations at query time).

Of course, there are use cases, where you'd still want to work
with TZ values in the database, so an extra converter for this
sounds like a good plan.
History
Date User Action Args
2017-01-11 17:16:25lemburgsetrecipients: + lemburg, ghaering, belopolsky, serhiy.storchaka, xiang.zhang, bozo.kopic
2017-01-11 17:16:25lemburglinkissue29099 messages
2017-01-11 17:16:25lemburgcreate