On Friday 22 September 2006 10:06, Christoph Zwerschke wrote:
> Sean Davis wrote:
>  > That is good to know.  Without testing it, does that mean that a
>  > db._pgdbCns__cnx is a pg connection object, or is it something else
>  > altogether?
>
> Yes, it's a pg connection object. Both pg and pgdb resort to the same
> underlying _pg module actually providing these connection objects.
>
>  > Thanks for the hint.  I'm relatively new to python and pygresql, after
>  > coming from perl.  Is there a place where the _pgdbCns__cnx is
>  > documented, or is it just a question of looking at the source?
>
> No, this is an undocumented feature. Actually it is a private attribute,
> so it seems D'Arcy wanted to discourage using it, probably because mixed
> use of both interfaces may lead to problems. The official way would be
> to open a separate connection using the pg module.
>
> However, I don't see a problem using the inserttable method that way.
> Should we consider a semi-official method of accessing the underlying
> _pg connection or even taking over all or some of its methods to the
> pgdb connection? D'Arcy, what do you think?

Thanks Christoph, for clarifying.

It seems to me that access to the pg-api methods would be considered an 
"add-on" feature, which I think the DB-API actually allows.  Would it make 
sense to make the non-redundant subset of the pg-api methods (those not 
included in the DB-API such as the database introspection methods and the 
socket connection methods) available via a public interface?  

Sean
_______________________________________________
PyGreSQL mailing list
[email protected]
http://mailman.vex.net/mailman/listinfo/pygresql

Reply via email to