Problem with database under Win32

[l3nz]

O wise brothers,
I have came across a couple of problems using Ado with Perl under Win32 connecting to a Ms SqlServer 7.0 database. I have no problems when executing a single Sql query but in a case I have to compute variables in the same query being sento to the database engine, like

declare @VAR int;
set @VAR = (select max(id) from table);
select * from table2 where id = @VAR

If I do this, the resulting recordset object shows strange behaviours, like

• it does not test positive as empty (i.e. when both $rs->Bof and $rs->Eof true) so that my code tries to loop on it; nevertheless, the loop is infinite because $rs->Eof is never true even if $rs->MoveNext reports no errors and gets executed some thousand times before the Ctrl-C
• I read somewhere that it is possible to access the different outcomes from each Sql statement separately but I do not know how to do it (in fact I'd like to access only the result from the third statement, as the first two are not supposed to return anything, and I wonder if that's the problem with the point above)
• I casually noticed that loading the Data::Dumper module seems to break the ability for the Ado connection to open a recordset (it just freezes)

Of course the query I'm using is by far more complex than the one I'm reporting here so replacing variables with subqueries is not a viable option.

Anybody got an idea on how to overcome this result? I cannot use DBI because it would mean a major rewrite of existing code. Thank you.

[NetWallah]

I would try one of two approaches:

1. Do your SQL in a stored procedure

2. Separate the 2 queries. Get the result of the MAX into a perl variable, then build and run the second query.

[l3nz]

Thanks for your reply - but how do I call a stored procedure on SQL Server from Perl?

[runrig]

Depending on the uniqueness of the id field in table1, and how close your real problem is to this example, but maybe this would work:

select table2.*
from table1, table2
where table1.id = table2.id
order by id desc

And fetch until id changes. And I don't know if DBI would help either, but if you have to rewrite the bits that don't work anyway, and if DBI does work, I'd consider that as an option. DBD::Sybase supports multiple result sets (and it supports SQL Server to some extent), I don't know if that is what you are referring to, or if that's applicable here.

[NetWallah]

Using SELECT TOP 1 From ..... ORDER BY ID DESC may be faster than trying a "Max".

I am not a SQL weenie though - you may want to check with your DBA as to why the subquery seems "slower".

[l3nz]

I am not sure - if you MAX() an index, you usually get top performance with no disk access at all; in any other case you need a table scan to know all the values, so you can max() them. My actual query is not any way similar to the one I am presenting here (it's a five-page belemoth, just in the case you wonder, even with all repeated subqueries extracted!)

[NetWallah]

Oh - here is a third, and perhaps simpler option - Combine the queries to get what you want like this :

select * from table2 where id =(select max(id) from table)

Untested (Because I don't have access to your database), but I have used SQL syntax similar to this.

[l3nz]

This is a subquery approach - I used it first, but query resolution gets much slower and query in itself gets way messier.

[l3nz]

I have fixed the problem - you cand find a solution (and new problems...) in the question ADO, multiple statements and Perl