Re: COPY support survey - Mailing list pgsql-jdbc

I would vote for 3.I already have ported your patch over to the
(near) current code and it is being used by a few people and from
what I understand quite stable. If you want I can fix this up and
commit it.

Dave
On 19-Aug-05, at 3:03 AM, Kris Jurka wrote:

>
> On and off there's been discussions of getting COPY support into
> the JDBC driver and I produced a patch to provide a stream based
> interface over a year and a half ago [1].  At the time we did not
> commit that because we had grand plans to provide a friendly
> interface to users so they did not have to know the internals of
> how COPY actually represents data.  This mostly happened off-list
> between Oliver and I, but the key suggestion is available here
> [2].  This allows you to use SQLOutput [3] calls like
> writeTimestamp or writeArray to directly send Java objects to the
> database without having to know the Postgresql specific text
> representation of them.  This turned out to be harder and more
> involved than it looked and I just let the patch languish, always
> intending to return to it, but as time has passed I've actually had
> less time to dedicate to the driver instead of more.
>
> I know a number of people are already/still using the original
> patch I provided, but at the same time they have also (indirectly)
> indicated that some additional support is needed [4].
>
> So the question is what should we do?
>
> 1) Just get the raw stream interface into the official release and
> let the end user deal with reading and writing the stream.  That's
> all I'll use and I don't care about anything else.
>
> 2) Don't give up, get the friendly SQLInput/SQLOutput interface in
> because this is what I really want.
>
> 3) Put in the raw stream interface, but keep working to also add
> the friendly one.  I'll understand if the interface changes and I
> have to modify my code a bit.
>
> Kris Jurka
>
> [1] http://archives.postgresql.org/pgsql-jdbc/2003-12/msg00186.php
> [2] http://www.ejurka.com/pgsql/copy/doc/sqlinputoutput.txt
> [3] http://java.sun.com/j2se/1.4.2/docs/api/java/sql/SQLOutput.html
> [4] http://archives.postgresql.org/pgsql-jdbc/2004-06/msg00165.php
>
> ---------------------------(end of
> broadcast)---------------------------
> TIP 9: In versions below 8.0, the planner will ignore your desire to
>       choose an index scan if your joining column's datatypes do not
>       match
>
>