SELECT ( * -City ) FROM Person.Address

If you need to select 99 of 100 columns in a table, you have to list all 99.  This can be a painful task.  For now, the best way to accomplish this is to generate a list of field names from the table, and copy them into your query.  Here’s how to do it.

Every database in SQL Server 2008 contains system views named sys.obects and sys.columns.  Sys.objects contains all database objects: system tables, user tables, constraints, views, etc.  Sys.columns contains all columns from every table, along with the same object_id of it’s parent object (the table).  By joining these system tables together, you can get a quick list of columns that you can copy and paste into your query:

USE AdventureWorks
GO
SELECT sc.name ColumnName
FROM sys.objects so
JOIN sys.columns sc
ON so.object_id = sc.object_id
WHERE so.name = 'Contact'

This is certainly one query to memorize!

To list columns from ANY user table, drop the table specification, add a filter on sys.objects for ‘type’, and include the table name by using the OBJECT_NAME function:

USE AdventureWorks
GO
SELECT OBJECT_NAME(sc.OBJECT_ID) TableName, sc.name ColumnName
FROM sys.objects so
JOIN sys.columns sc
ON so.object_id = sc.object_id
WHERE so.type = 'U'

This query can be taken even further by joining to the sys.types view, where data type metadata is stored:

USE AdventureWorks
GO
SELECT OBJECT_NAME(sc.OBJECT_ID) TableName, sc.name ColumnName,
t.name [DataType], sc.is_nullable Nullable, sc.is_identity Is_identity
FROM sys.objects so
JOIN sys.columns sc
ON so.object_id = sc.object_id
JOIN sys.types t
ON sc.user_type_id=t.user_type_id
WHERE so.type = 'U'

To apply granular permissions to a specific object, like a table, use the GRANT [permission] ON [object] TO [user] format.

“Increasing the value of min memory per query may improve performance for some small to medium-sized queries, but doing so could lead to increased competition for memory resources.”

Updates to the ‘min memory per query (KB)’ setting can be done like this:

--first, turn on advanced options
SP_CONFIGURE 'show advanced options', 1
GO
RECONFIGURE
GO
--then, set new value - default value is 1024KB, this increases it to 2MB
SP_CONFIGURE 'min memory per query (KB)',2048
GO
RECONFIGURE WITH OVERRIDE
GO
--now, display new value
SP_CONFIGURE 'min memory per query (KB)'
GO
RECONFIGURE WITH OVERRIDE
GO
___________________________________________________________________________________________________

http://msdn.microsoft.com/en-us/library/ms181047.aspx

SELECT * INTO new_tbl
FROM old_tbl

This will create a copy of the table, data and all.  What if you don’t want the data copied with the schema?  My friend Alessandro Basso uses this method:

SELECT * INTO new_tbl
FROM old_tbl
WHERE 1=2

To play with this, let’s create some sample data:

Now we have a table with sample data:

Using the data we just created, we want to show only the most recent transaction for EACH customer.   We can do a simple GROUP BY, but if we include TransactionID in the result set, we are forced to get a record for each TransactionID, since it has to be contained in the GROUP BY’s list of non-aggregates:

We get the full result set – not what we want.  To include TransactionID (or any other field that may be in the table) in the result set, but grouping only on CustID, we can include the ROW_NUMBER function in the select list, using the PARTITION BY clause:

We get this:

Using the ROW_NUMBER function in the select list (with out the PARTITION BY parameter) simply gives us the unique row number for each record.  When we use the PARTITION BY clause, it groups on CustID.  The ORDER BY clause is required, and necessary for our results – this allows us to make the most recent SaleDate = row 1.

I give credit to Dan Humphries for showing examples of ROW_NUMBER here.

http://msdn.microsoft.com/en-us/library/ms186734.aspx

To change the UAC setting, click the Start button type ‘UAC’ in the Search box. Open the UAC and change to the desired setting.

http://technet.microsoft.com/en-us/library/bb742535.aspx

http://www.sql-server-performance.com/tips/composite_indexes_p1.aspx

When I first encountered the effects of this setting, I was attempting to rearrange the column order of a table. Using the GUI, I went to the table design view and moved one column in front of another.

When I hit ‘Save’, I received this message:

When a table’s schema is changed, the process that SQL Server performs drops the table and recreates it. The new setting that prevents this is called ‘Prevent saving changes that require table re-creation‘, and it’s easy enough to turn off. I’m sure it has been put in place as a safeguard to prevent unintended table changes. To turn off the setting, go to (on the menu bar) Tools/Options/Designers/Table and Database Designers, then under the Table Options section, uncheck Prevent saving changes….

Schema changes will not be blocked.  You’re welcome.

To demonstrate how SQL Server can solve this problem, I’m going to use the AdventureWorks database and the Production.Product table. This table has a Color field. I want to find (in one query) the number of records for each respective color (or no color: NULL). Here’s a quick shot of part of the table:

I can see that there are the colors Black and Silver in the table, along with many NULLs. To get the count of records having the color Black, I can run the following:

That gives me a count of 93 records. But how many different colors are there? To find out, I’ll get all the DISTINCT Color values:

NOW I can formulate my entire query accurately, getting a total count, NULLS count, and counts for every color:

Looks good!

Access IIF statement
CASE statement