[okfn-help] vdm: strange behavior

[Rufus Pollock]

On 7 September 2010 19:56, Martijn Faassen <faassen at startifact.com> wrote:
> Hi there,
>
> On Tue, Sep 7, 2010 at 7:56 PM, Rufus Pollock <rufus.pollock at okfn.org> wrote:
>>> It might be that the exact table structure we create depends on the
>>> database system in use; PostgreSQL could use DateTime and MySQL could
>>> use a float. This might lead to hard to understand code however.
>>> Alternatively we could translate the float column back to DateTime
>>> objects on the fly, with a read-only property. But it would still make
>>> direct queries of the revision table more difficult.
>>
>> You could implement this at a lower level in sqlalchemy, i.e. by
>> creating a custom type. In many ways if we are going to support mysql
>> (despite it's poor timestamp behaviour) I would like to special case
>> it where possible and not disturb the existing db structure (for
>> psql/sqlite etc) -- even if this is a bit ugly!
>
> I'm not sure I'm not reading this out of order, but I've reported
> already a while ago that I created a special type for SQLAlchemy that
> stores a higher-precision datetime as a decimal.

But won't this require migrations for all systems currently using VDM
in that you are changing the underlying type (i.e. from datetime to a
decimal).

> I think this is the right way to go, as at least at the developer
> level the behavior of that column will be the same as for postgresql,
> as long as the SQLAlchemy APIs are used.

OK, so there is no change to postgres column type involved here?
That's great news!

Rufus

> Regards,
>
> Martijn
>



-- 
Open Knowledge Foundation
Promoting Open Knowledge in a Digital Age
http://www.okfn.org/ - http://blog.okfn.org/