Re: SQL question

[Vern Hamberg <vhamberg@xxxxxxxxxxx>]

David

I don't think it is "bad" as such. But it might be slower. With F-specs 
you are using record-level access - you go directly to the records, it 
is very efficient. With SQL, the system has to figure out just HOW to 
get the data - with F-specs, YOU decide already. See, with this 
statement, you have dynamic SQL. Although it is embedded in an RPG 
program, it is not static. IBM have made dynamic SQL better and faster, 
but no guarantees.

You would have to run some tests to see what the difference is in timing.

It also might not matter which you use, especially if this is an 
interactive program and the statement is not executed very often. If it 
is in a batch process over thousands and millions of records, it might 
matter which one you use.

Now just so you know - MYTABLE1 and MYTABLE2 should be physical files, 
not logical files. When you specify logical files (indexes in other SQL 
products), the optimizer has to take the time to figure out which 
physical file the LF is based on, then continue to work out the best LF 
to use, if any. It is not just going to use what you thought you told it 
to use, unlike record-level access - there's that expression again!!

HTH
Vern

David FOXWELL wrote:
> Hi,
>
> If I understood correctly from a previous thread, the use of SQL in the example below taken from something I already wrote is not a good idea as I am doing a single row search.
>
> EXEC SQL
>
> SET : MyField = (
>
> SELECT MYFIELD FROM MYTABLE1 JOIN MYTABLE2 ON ID1 =ID2
>
> WHERE MYOTHERFIELD= : MYSEARCHDATA );
>
> If I were not to use SQL I'd have to code the 2 files in the F specs, and do 2 CHAINs. I've noticed a lot of people are doing similar quick fixes.
>
> Sorry to keep asking the same question, but is this example really that bad?
>
>
>