Link to home
Start Free TrialLog in
Avatar of Iver Erling Arva
Iver Erling ArvaFlag for Norway

asked on

What is wrong?

I have this query:
declare
  type nt_type is table of string(10);
  nt nt_type := nt_type ('08.05.2013','09.05.2013');
begin

for i in 1..nt.count loop

select request_id, request_emode, request_pgm, pp.description, name_id_no, substr(addressee,1,15), policy_no, pr.print_job,request_date 
  from print_request pr, print_program pp
where request_pgm = program_id
  and request_pgm not like ('B8350')
  and request_pgm not like ('BAT%')
  and request_pgm not like ('F%')
  and request_pgm not like ('KX%')
  and request_pgm not like ('269')
  and (pr.printer_id in (1,2) or pr.document_pile is not null)
  and request_emode not like ('X%')
  and trunc(request_date) = to_date(nt(i), 'DD.MM.YYYY')
order by pr.print_job desc, name_id_no;

end loop;
end;

Open in new window


If I run the select statement and replace nt(i) with e.g. '08.05.2013' it works fine, but running it like this gives med the error:
Error report:
ORA-06550: linje 8, kolonne 1:
PLS-00428: an INTO clause is expected in this SELECT statement
06550. 00000 -  "line %s, column %s:\n%s"
*Cause:    Usually a PL/SQL compilation error.

Open in new window


What could be the problem? All I want to do is to run the same select several times but with different date values.

Thx!
Avatar of Sean Stuber
Sean Stuber

>>> replace nt(i) with e.g. '08.05.2013'


Ff that works, then you're seeing an oracle bug.
A select without an INTO clause is wrong whether you're using variable or literals.

In other words, the error message is correct for the code you have posted


When you run the select,  what are you expecting to do with the results?
That's why you need an INTO, so the sql output has somewhere to go
I agree, an implicit cursor (like yours) in PL\SQL requires an "into" line with either a record variable or discrete variables for each column being selected.

I see another potential problem in your query, that is: "trunc(request_date)".  In a tiny test system, this will not be a problem.  Also, in a new production system with few (or no) records initially, this will not be a problem either.  But, as the number of records in this table increases, and especially if this column is indexed, your "trunc" will prevent the use of the index and force Oracle to read every record that meets the other criteria.  That could add a huge performance penalty as the data in this table increases.

Try to avoid using any operators like: trunc, upper, lower, to_char, to_date, etc. on database columns referenced in your "where" clause.  Do the conversion on the other side of the '=' sign instead.  For date columns that may include a time-of-day, sometimes that means you will have to use "between" instead of "=", something like this:

and request_date between to_date(nt(i), 'DD.MM.YYYY') and to_date(nt(i), 'DD.MM.YYYY') + 86399/86400

Explanation:
1. There are 86,400 seconds in a day (60 x 60 x 24).
2. Your first "to_date" with no time component included, will mean a time of midnight (the first second in the day).
3. The second "to_date" adds enough seconds to get to the last second of the day, so a "request_date" value that includes any time of that day will be included..
Avatar of Iver Erling Arva

ASKER

First you ask what I expect, I'd expect the output of two SELECTs in the output window of Oracle SQL Developer just as I get the output from one select when I enter it and run it.

I will try your suggested solutions tomorrow. Partly this is also since I'm trying to learn.
I have used a lot of relatively straight forward SELECTs for many years, but have used another programming language to manipulate and get the results that I wanted. I thought it would be interresting to go one step further and do it all using sql.

I hear what you say about the INTO, but I have never used it like this. My selects usually are in the form:
SELECT column(s) FROM table(s) WHERE condition(s) ORDER BY column(s).

Open in new window

That's it. It has always worked fine.
But if I have to use INTO in this case, I would like to know what to use to get it to print the output columns in the output window of SQL Developer.
Thanks!
When you enter a select statement in the editor window of sql developer your execution environment is sql developer and the sql engine returns the output to that.

When you run it as you have shown above your environment is your pl/sql  block and the sql engine returns the output to that and that's why you need an "INTO" to give the output someplace to go.

sql developer has an output grid,  pl/sql does not.
Ok, so what should I enter to send the output to SQL developer then? Is that possible?

Thx!
ASKER CERTIFIED SOLUTION
Avatar of Mark Geerlings
Mark Geerlings
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Back to where I started, then...