[Pgpool-general] Multi-statement queries

[Matthias Tief]

Thank you for your rapid answer.

But for me it is still not clear. What you describe seems to be a
collection of statements transmitted as a single transaction. But this
is done very well by pgpool-II in my tests. I am wrong with that?

So than my question is: What is the difference between "multi-statement
queries" and a single transaction of multiple statements?

As I understand pgadmin implicitly wrapps a BEGIN/COMMIT block around
all statements, that it becomes a single transaction.

Looking forward to your answer.

Matthias

Guillaume Lelarge schrieb:
> Le 05/03/2010 09:45, Matthias Tief a écrit :
>> Hallo,
>>
>> we are planning to use pgpool-II as replication solution for our
>> PostgeSQL application. Currently we evaluate, whether the restriction of
>> pgpool-II affect our application. In the documentation you say:
>> "pgpool-II cannot process multi-statement query." What do you exactly
>> mean with "multi-statement queries". As I found out you don't mean
>> transactions that contain multiple statments, like BEGIN - COMMIT
>> blocks. These blocks are "committed" or in case of a failure are "rolled
>> back" correctly. I would appreciate if you could give me an example of
>> multi-statement queries.
>>
> 
> "statement1; statement2; statement3;" in a single trip to the backend.
> 
> For example:
> 
> "INSERT INTO t1 VALUES (1); INSERT INTO t1 VALUES (2); INSERT INTO t1
> VALUES (3);"
> 
> is a multi-statement query, whereas
> 
> "INSERT INTO t1 VALUES (1);"
> then
> "INSERT INTO t1 VALUES (2);"
> then
> "INSERT INTO t1 VALUES (3);"
> 
> are three one-statement queries.
> 
> pgAdmin uses this quite a lot ("SET client_encoding TO utf8; SELECT *
> FROM...").
> 
> 

-- 
IT Service Omikron GmbH
Mohrenstr. 63-64; D-10117 Berlin
Tel.: +49 (030) 220791-30
Fax: +49 (030) 220791-55
Email: matthias.tief at itso-berlin.de
Internet: www.itso.de
Amtsgericht Berlin Charlottenburg HRB 21013