On Sun, Jan 20, 2002 at 08:00:33PM -0800, Sterin, Ilya wrote: > > > > -----Original Message----- > > From: Tim Bunce [mailto:[EMAIL PROTECTED]] > > Sent: Sunday, January 20, 2002 2:16 PM > > To: Sterin, Ilya > > Cc: [EMAIL PROTECTED] > > Subject: Re: bind_param() > > > > > > It would have to be a new method. Perhaps > > New method is fine, I don't see a problem with it. Though I thought > about just differentiating between refs and vals in the bind_param > version, therefore this will make it kind of a seemless interface. I'm > not sure about underlying obstacles for this though.
The main one is that drivers should be free to accept and return values that are implemented as refs to objects etc if they want to. > Maybe I'm going a little overboard here, but something that will allow > both types, would be nice as well. Possibly some values will change and > some won't, therefore the binding of the ref will be delayed, but the > values can be bound right away? > > What do you think? Sure, it would be per-param. > > bind_param_at_execute(...) > > bind_param_lazy(...) > > bind_param_ref(...) > > > > The driver could just defer the actual bind till execute is > > called. I might just be able to get the Driver.xst template > > code to handle that automatically. > > I wish I would have some time to look at it. If I do sometime next > week, I'll let you know and probably ask a >>few<< questions as well:-) Sure. Ask _before_ you satrt coding though :) Tim.
