July 8, 2014 at 7:26 pm
Comments posted to this topic are about the item Stairway to T-SQL: Beyond The Basics Level 9: Dynamic T-SQL Code
Gregory A. Larsen, MVP
July 23, 2014 at 4:38 am
Hi Greg,
I do not go quite with your conclusions.
If parameterized SQL is used as it is intended, the skeleton of the statement @CMD provided by the application and user input used only for parameters,
there will be no danger of SQL injection and consequently no advantage in avoiding dynamic SQL.
Concluding from what I know and what is stated in the article.
reagrds
Herbert
July 23, 2014 at 8:41 am
Hello Greg,
Another good article. I have enjoyed this stairway series.
In this article I noticed a few inconsistencies that may be from changing your examples?
Several places you use "DELETE" where maybe it should be "DROP" or "DROP TABLE"?
You refer in some captions to a stored procedure called "GetUserName" and in the text to an @EMAIL parameter
Also it looks like this sentence: "In this example I deleted the Client table." should read "In this example I dropped the Product table."
July 23, 2014 at 4:54 pm
Greg, thank you for article.
Could you please refrain from providing SELECT * FROM table example - it's a proverbial ticking "time bomb".
Changes in table schema like creating new or dropping existing column could trigger a chain reaction of bug further down stream in middle-tier and/or on UI. My favorite anti-pattern - combining SELECT * with querying system views EVERY time somebody try to retrieve data. in this scenario even renaming a column could be a potential problem.
July 24, 2014 at 7:38 am
Do the cursors in the example of looping dynamic sql act as an implicit go statement?
July 24, 2014 at 8:42 am
Robert.Sterbal (7/24/2014)
Do the cursors in the example of looping dynamic sql act as an implicit go statement?
Yes, unless it is wrapped in an explicit transaction.
😎
July 24, 2014 at 8:48 am
Thanks Gregory for this nice peace, good job!
Missed seeing one essential thing though in the injection part and that is a strong coupling to the metadata. If a user can select in any form the objects (table/view/procedure etc.) to use, one should do a firm check for it's existence. A quick sample would be
😎
DECLARE @SCHEMA NVARCHAR(128) = N'dbo';
DECLARE @TABLE NVARCHAR(128) = N'CLIENT';
DECLARE @COLUMN NVARCHAR(128) = N'CLIENT_NAME';
DECLARE @SQL_STR NVARCHAR(MAX) = N''
SELECT
@SQL_STR = N'SELECT ' +
COLUMN_NAME
+ N' FROM ' + TABLE_SCHEMA
+ NCHAR(46) + TABLE_NAME
FROM INFORMATION_SCHEMA.COLUMNS
WHERE COLUMN_NAME = @COLUMN
AND TABLE_NAME = @TABLE
AND TABLE_SCHEMA = @SCHEMA
--- and so on
July 24, 2014 at 12:17 pm
To prevent SQL Injection when the names of objects or columns are entered by the user, I'd recommend as well QUOTENAME() function. Of course, that doesn't remove the good practice of using parametrized queries. Combining both options, you should be safe. If anyone prove me wrong, we can all learn some more.
DECLARE @Table varchar(128) = 'sys.tables; '' SELECT TOP 100* FROM sys.columns'
DECLARE @sql nvarchar(4000)
SET @sql = 'SELECT * FROM ' + QUOTENAME(@Table)
PRINT @sql
EXEC sp_executesql @sql
July 29, 2016 at 7:40 am
Thanks for another great article.
August 1, 2016 at 6:48 am
Good article. We just try to avoid it altogether. We have some and are in the process of getting rid of it.
Viewing 10 posts - 1 through 9 (of 9 total)
You must be logged in to reply to this topic. Login to reply