RE: Embedded SQL with Coalesce

Vern and Chuck,

Hate to see you both spending any more energy on this.
To quote an earlier reply:
The line you cite was intended to show the difference between the one that ran in iNav with 2011 and the embedded SQL with :RPTTXYR - a program variable.
Never was intended to concatenate the two. Sorry for the trouble.

HTH,
TIA,
Thanks,
Whichever applies

Dave B

"It's amazing what you can accomplish if you don't care who gets the credit."
Harry S. Truman

-----Original Message-----
From: rpg400-l-bounces@xxxxxxxxxxxx [mailto:rpg400-l-bounces@xxxxxxxxxxxx] On Behalf Of Vern Hamberg
Sent: Wednesday, May 30, 2012 2:39 PM
To: RPG programming on the IBM i / System i
Subject: Re: Embedded SQL with Coalesce

Then, Chuck, maybe the issue is using scalar functions on a column you are using to test - like

where substring(somecolumn, 5, 4) = '1234'

That, IIRC, would prevent use of an index. I think!

Vern

On 5/30/2012 1:26 PM, CRPence wrote:

The host variable has a value that is constant for the scope of
the statement. Thus there would be no reason to calculate the
expression more than once. The expression (2011 concat :RPTTXYR) can
be evaluated [as cast to the data type\attributes of the column being
compared; implicit vs explicit CAST should be moot] and implemented as
the effective literal it is, before the rest of the query
evaluation\optimization begins. That expression, the effective
constant, just as a literal would is constant, is a good potential
candidate for implementation on an index defined on column ARTXYR when
coded in the WHERE clause as:
and A.artxyr = 2011 concat :RPTTXYR

While it is entirely possible the optimizer does not have the
capability to do so [I believe it does], the query implementation
_should_ be capable of effecting the equivalent of the following for
its optimization _because_ the given expression is as good as a literal:
and A.artxyr = :Evaluated_Expression_2011_concat_hvRPTTXYR

I would expect VE to confirm that potential.

Regards, Chuck

On 30 May 2012 09:50, Vern Hamberg wrote:

Tom - I'll jump in here - the calculation has to be done for every
record you are testing - this prevents the use of any index on
A.artxyr.

This is generally true, and may depend on where the calculation is.

Also, I'm inclined to think implicit casts may have some bearing on
performance. Not sure, would have to run through Visual Explain to
see.

In this case, it'd be easy enough to run 2 SELECTs in Navigator, one
with this concatenation of 2011 and a constant that is a valid value
for RPTTXYR. The other would use directly the value that resulted
from the concatenation.

On 5/30/2012 10:06 AM, Tom E Stieger wrote:

On a side note, why does this matter for performance?

Birgitta Hauser on Tuesday, May 29, 2012 10:38 PM wrote:
Scott Klement on Tuesday, 29.5 2012 23:58 wrote:

This line looks suspicious to me:
and A.artxyr = 2011 || :RPTTXYR

To me too!
For performance issues you should avoid concatenating those things
in the SQL statement, calculate the variable value in your RPG and
use it as follows:

MyVar = '2011' + RPTTXYR;

Select ...
Where a.ARTXYR = :MyVar

--
This is the RPG programming on the IBM i / System i (RPG400-L) mailing list To post a message email: RPG400-L@xxxxxxxxxxxx To subscribe, unsubscribe, or change list options,
visit: http://lists.midrange.com/mailman/listinfo/rpg400-l
or email: RPG400-L-request@xxxxxxxxxxxx
Before posting, please take a moment to review the archives at http://archive.midrange.com/rpg400-l.