> Mark Gibson wrote:
>
> I've found that dblink works without the oid map, ie: dblink(... , ...
> , false)
> but returns the error when enabled, ie: dblink(... , ... , true)
>
Hello, I've found the problem, although I'm still a bit confused by it.
When the call to SPI_finish() at the end of pgresultGetTupleDesc
in my patch is removed, it works fine, custom datatypes and all :)
But I still don't understand why this is. Is this safe?
Also, I was thinking about how we could uniquely identify a database.
Are there any variables we could retreive from the remote db after
connecting to it that could form a consistent unique id for that db?
Is there a way we can determine whether 'pg_type'
in the remote database has been modified?
We could automatically update the oid map in that case.
--
Mark Gibson <gibsonm |AT| cromwell |DOT| co |DOT| uk>
Web Developer & Database Admin
Cromwell Tools Ltd.
Leicester, England.