[TYPO3-dev] RFC: Change roadmap for 4.5 and 5.0

[Dimitri Tarassenko]

Elmar,

On 4/11/06, Elmar Hinz <elmar.DOT.hinz at team.minus.red.dot.net> wrote:
> Dimitri Tarassenko schrieb:
> > - tslib_db and DBAL should die and be replaced with ADOdb-based layer.
> > Along this way, dates/times in the DB should BE dates and times, not
> > Unix timestamps.
>
> When I look at the class name chaos of ADO I wonder if there is no better
> alternative. In my eys the classnames of ADO are risky pollution of any
> namespace. Not to ask for PEAR standards at all.

True. However ADOdb is currently the only abstraction level framework
written in PHP that has decent overhead, and it is probably the one
most used by developers. I do think it's a viable trade-off,
especially considering that DBAL is already built on ADOdb and adodb
is a system extension now.

> Unix timestamps are intergers. They are a standard since the a standard
> since the 70s that can't even change. Dates and times in databases could
> rather tend out as a point of incompatibility and misanderstandings. If I
> only think of all timezone, localization formats, summer/winter time.
> Changing UNIX timestamps against something like that is the road to Babylon.

Elmar, I've been around;))) I do know that they are and why they are
there, but following your logics why do you think SQL developers
bothered creating date types at all? Try for example writing a select
statement to pick Tuesdays out of integers containing unix timestamps,
and you'll end up converting them to date types anyway. ADOdb is there
precisely to smooth all the troubles with date formats.

--
Dimitri Tarassenko