Re: contrib: hack for psycopg2 performance

On Thu, May 13, 2010 at 01:47:42PM +0300, Marko Kreen wrote:

> > -1 for breaking existing features in a module that's been in use for
> >  some 10 years. I wouldn't do it for introducing an useful feature such
> >  as nextset(), even less to simplify an implementation.
> 
> In that case, separate argument to connection or cursor constructor?
> 
> And instead of toggling PQexecParams always-on, I'd suggest using
> the above logic, so it would be usable in more situations.
> 
> Then old code keeps working and new code can default to better
> argument handling.

That's what I've been arguing all along but it's been
disliked so far.

Karsten
-- 
GPG key ID E4071346 @ wwwkeys.pgp.net
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346