Question

We have a web application in which the users perform ad-hoc queries based on parameters that have been entered. I might also mention that response time is of high importance to the users.

The web page dynamically construct a SQL to execute based on the parameters entered. For example, if the user enters “1” for “Business Unit” we construct a SQL like this:

SELECT * FROM FACT WHERE 
BUSINESS_UNIT = '1'
--AND other criteria based on the input params

I found that where the user does not specify a BUSINESS_UNIT the following query is constructed

SELECT * FROM FACT WHERE 
BUSINESS_UNIT LIKE '%'
--AND other criteria based on the input params

IMHO, this is unnecessarily (if not grossly) inefficient and warrants sending the code bad for modification but since I have a much higher rate of sending code back for rework than others, I believe that I may be earning a reputation as being “too picky.”

If this is an inappropriate question because it is not a direct coding Q, let me know and I will delete it immediately. I am very confused whether subjective questions like this are allowed or not! I’ll be watching your replies.

ty

---

Update:

I am using an Oracle database.

My impression is that Oracle does not optimize “LIKE ‘%'” by removing the condition and that leaving it in is less efficient. Could someone confirm?

Answer 1

Although that looks grossly inefficient, I just tested it out in SQL Server, and the query optimizer was smart enough to filter it out.

In other words,

SELECT * FROM FACT WHERE 
BUSINESS_UNIT LIKE '%'

and

SELECT * FROM FACT

generated the exact same query plan. So there shouldn’t be a performance difference (depending on your DB engine I guess), although it does look kind of sloppy.

That may affect your decision to send it back or not. Personally I probably would, but if you’re under a cloud already then it’s something you can probably relax on, at least in terms of performance.