Re: [cx-oracle-users] type object matching - BINARY

[Anthony T. <ant...@gm...>]

On 9/13/06, Mario Frasca <mf...@zo...> wrote:
> On 2006-0913 11:33:04, Anthony Tuininga wrote:
> > [...] I won't be adding anything along the lines of
> > DBAPITypeObject() as it doesn't make any sense -- see above. If
> > that is something you really care about you can always create
> > your own Python wrapper.... :-)
>
> pity, I don't like needing wrappers, I prefer using things out of the box
> (sounds like the classical lazy developer), even if that involves needing
> to patch them.  I also don't like the style chosen in the db-api2:
> testing for equality and implementing it as a test for inclusion is
> quite confusing, to say the least.  subclassing seems more appropriate.
> as far as I understood, it has been implemented like this because this
> way you can define a CLOB to fall into a BINARY but also into a STRING...
> you can achieve this also by multiple inheritance, but maybe it has been
> considered overkill, less efficient, I don't know, just making hypoteses.

I don't like wrappers either -- but the great goal of being able to
simply drop in a database driver just isn't there. There are too many
discrepancies and the DB-API does not cover them (and doesn't really
try to, either). Its been great for making the interfaces __similar__
-- and that's good enough for me, for now.

> I had seen there was some movement on db-sig and a possible new version
> of the db-api agreements.  until then, my library will have to cope with
> so many different interpretations of the directives...  hope this better
> db-api3 comes soon!

Agreed.

> greetings,
> Mario