I am not disagreeing about what the user wanted, and yes it is nit
picking, but my feeling is that the DB guys have been at this alot
longer than EJB-QL and EJB-QL should be the object version of SQL99. If
that is the case either you modify the EJB-QL spec of say no you can not
do this because SQL9
User writes "ORDER BY y" and gets something not ordered by y, but rather by
the DISTINCT's selection order on x, which can change from call to call
based on activity on the tab table. That's what the user will get in the
situation that x's map to multiple y's.
It's may be obvious what the user wa
Dave that is not what I said. In the case that there happens to be two
rows with the same x value and different y values, I don't care.
Otherwise I do.
Bases on what you have posted, how do you think we should handle this
case in EJB land? Should we just throw an error saying too bad? One
o
t: Monday, February 10, 2003 6:24 PM
Subject: Re: [JBoss-user] JBoss Generating Some Bad SQL
> If you do not care how it is ordered then why bother with the order by
> clause? The database is correct , you are not.
>
>
> Dain Sundstrom wrote:
>
> >
> > On Mond
If you do not care how it is ordered then why bother with the order by
clause? The database is correct , you are not.
Dain Sundstrom wrote:
On Monday, February 10, 2003, at 01:53 PM, Dave Smith wrote:
No, it isn't. Consider
SELECT DISTINCT x FROM tab ORDER BY y;
Assuming there
On Monday, February 10, 2003, at 01:53 PM, Dave Smith wrote:
No, it isn't. Consider
SELECT DISTINCT x FROM tab ORDER BY y;
Assuming there are multiple values of y for any given x, how would you
expect the result to be sorted? It's ill-defined.
There is nothing ill defined by this.
My request ..
For this query ..
SELECT DISTINCT t0_s.rec_num FROM sample_request t0_s,
sample_request_line_item t1_li WHERE (t1_li.fulfillment_status =
'Shipped'
AND t1_li.followup_date <= '2003-02-04 20:00:00.0+00' AND
(t1_li.customer_prospect_feedback IS NULL OR
t1_li.customer_prospec
Send me the schema's and I'll post a question on the pg lists, asking
why.
On Fri, 2003-02-07 at 16:11, Hunter Hillegas wrote:
> I am running my dev environment on JBoss 3.2 RC1...
>
> I am using JBossQL to override a query.
>
> My JBossQL statement is as follows:
>
> SELECT DISTINCT OBJECT(s)
date_of_request
I've experience with INFORMIX, and this select is wrong for INFORMIX.
vlk
-Original Message-
From: Dain Sundstrom [mailto:[EMAIL PROTECTED]]
Sent: 8. februara 2003 16:47
To: [EMAIL PROTECTED]
Subject: Re: [JBoss-user] JBoss Generating Some Bad SQL
We have the s
wrong for INFORMIX.
vlk
> -----Original Message-
> From: Dain Sundstrom [mailto:[EMAIL PROTECTED]]
> Sent: 8. februara 2003 16:47
> To: [EMAIL PROTECTED]
> Subject: Re: [JBoss-user] JBoss Generating Some Bad SQL
>
>
> We have the same problem
We have the same problem with Oracle. I personally think this is a bug
with both of the databases. I see no reason why the query processor
can't figure out the correct way to form a internal query object; we
can.
I'll ask Jeremy if he wants to work this out in in 3.2. Otherwise
you'll have
PostgreSQL does not allow an aggregated field reference to appear in the
where clause.
SELECT DISTINCT t0_s.rec_num ... WHERE ... t0_s.rec_num=t1 ...
Guess DISTINCT by itself is considered an aggregate rather than a
pre-aggregate filter. Possible solutions: a) fix PostgreSQL so that it does
not t
12 matches
Mail list logo
