Thursday, June 12, 2008

Re: [HACKERS] Options for protocol level cursors

On Jun 12, 2008, at 4:45 PM, Tom Lane wrote:
> Huh? I don't see why... you might have such a limitation in a
> particular driver, but not in the protocol.

Oh? I know when you bind a prepared statement you have the ability
state the formats of each column, but I'm not aware of the protocol's
capacity to reconfigure the formats of an already existing cursor; ie,
a DECLARE'd cursor. I know you can use the Describe message to learn
about the cursor's column types and formats....

Got a link to the part of the protocol docs describing this feature?

>> Also, the latter has other problems wrt statement parameters. I guess
>> you
>> could prepare(protocol level) the DECLARE, but that seems like a
>> gross
>> workaround as it defeats the purpose of prepared statements by
>> forcing
>> you
>> to create a new statement for each cursor that you plan to open.
>
> Well, using a query for a cursor is grounds for replanning anyway,
> because you might want a fast-start plan in such a case. And it's
> *definitely* grounds for replanning if you are asking for SCROLL
> capability --- the plan stored for a regular prepared statement
> very likely can't support that.

Ah, that is good to know. Thanks.

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

No comments: