Re: feature request

On Tue, Aug 30, 2011 at 8:30 PM, Guillaume Lelarge
<guillaume(at)lelarge(dot)info> wrote:
> On Tue, 2011-08-30 at 19:29 +0100, Dave Page wrote:
>> On Tue, Aug 30, 2011 at 6:15 PM, Guillaume Lelarge
>> <guillaume(at)lelarge(dot)info> wrote:
>> >>
>> >> libpq doesn't support multiple resultsets in a single query unfortunately.
>> >>
>> >
>> > As a matter of fact, it does (see
>> > http://pgolub.wordpress.com/2009/11/17/script-slicing-by-pgmdd/), and I
>> > really want pgAdmin to use this. Still need to find a good UI and some
>> > time to work on it :)
>>
>> Err, no it doesn't. The blog you've referenced describes how PgMDD
>> takes the PostgreSQL 8.3 parser and uses it to pre-parse scripts into
>> individual statements on the client, which it then executes
>> individually. That's similar to what's being asked for here, but has
>> the rather nasty limitation of not understanding PG 8.4+'s grammar.
>>
>> Libpq itself only supports a single resultset for multi-part
>> statements. Even in 9.1, the docs say:
>>
>> The command string can include multiple SQL commands (separated by
>> semicolons). Multiple queries sent in a single PQexec call are
>> processed in a single transaction, unless there are explicit
>> BEGIN/COMMIT commands included in the query string to divide it into
>> multiple transactions. Note however that the returned PGresult
>> structure describes only the result of the last command executed from
>> the string. Should one of the commands fail, processing of the string
>> stops with it and the returned PGresult describes the error condition.
>>
>> (http://www.postgresql.org/docs/9.1/static/libpq-exec.html#LIBPQ-EXEC-MAIN)
>>
>> In other words, only the last resultset is accessible.
>>
>
> You're right for PQexec. And wrong for PQsendQuery, which is the one we
> use on the query tool.
>
> And to quote the fine manual :)
> (http://www.postgresql.org/docs/9.1/static/libpq-async.html#LIBPQ-PQSENDQUERY):
> After successfully calling PQsendQuery, call PQgetResult one or more
> times to obtain the results. PQsendQuery cannot be called again (on the
> same connection) until PQgetResult has returned a null pointer,
> indicating that the command is done.

Oh, I stand corrected. That's pretty neat - I had read that before I
think I mis-parsed it to mean that you call PQgetResult to get more
results for the one and only resultset.

--
Dave Page
Blog: http://pgsnake.blogspot.com
Twitter: @pgsnake

EnterpriseDB UK: http://www.enterprisedb.com
The Enterprise PostgreSQL Company