On May 24, 2007, at 20:39 , Andrew Sullivan wrote:
> On Thu, May 24, 2007 at 03:25:52PM -0400, A.M. wrote:
>> Wouldn't it be a cool feature to persists transaction states
>> across connections so that a new connection could get access to a
>> sub-
>> transaction state?
>
> You could do this using an incredibly evil, carefully implemented
> hack in a connection pool. I'm shuddering at the thought of it, to
> be honest, so details are left as an exervisse for the reader.
Actually, a sample implementation could be done using stored
procedures and some IPC. It would however require that the receiver
poll for requests- the API would probably look very similar to dblink.
-M