Thursday, May 25, 2006

Populating A Dropdown With Weekly Dates

Let's assume you have a website and the user can pick from a date range, You want to be a nice guy/girl (* reason for the asterisk is at the end) so instead of having a dropdown that looks like

Week1
Week2
Week3

You want it to have start and end dates for the week
The start date should be a Sunday and the end date should be a Saturday
The output should be something like this

2006-05-07 - 2006-05-13
2006-05-14 - 2006-05-20
2006-05-21 - 2006-05-27

What do you do? Do you store all these dates in a table, you can but you will have to maintain this.
Another way is to use a number table
You would create the number table only once. Then you can use DATEADD and DATEPART with the number table to return the desired results


-- Create out Pivot table ** do this only once-- populate it with 1000 rows
CREATE TABLE NumberPivot (NumberID INT PRIMARY KEY)

DECLARE @intLoopCounter INT
SELECT @intLoopCounter =0

WHILE @intLoopCounter <=1000 BEGIN
INSERT INTO NumberPivot
VALUES (@intLoopCounter)

SELECT @intLoopCounter = @intLoopCounter +1
END
GO

--And here is the query

--I started with a Sunday
DECLARE @StartDate DATETIME
SELECT @StartDate ='05/07/2006'

--How many rows to return
DECLARE @RowsToReturn INT
SELECT @RowsToReturn = 50

SELECT DATEADD(wk,numberID,@StartDate) AS WeekStart,
DATEADD(dd,-1,(DATEADD(wk,numberID+1,@StartDate))) AS WeekEnd
FROM dbo.NumberPivot
WHERE NumberID <= @RowsToReturn -1 -- 1 since the number table starts from 0
AND DATEADD(wk,numberID,@StartDate) >=@StartDate
ORDER BY 1


* At my previous job we were doing file exchanges with some state agencies. One of the fields in the file was Sex and the values could be either F or M,. fair enough right? Well guess what, there was one person who worked there and had a sex-operation done and did not want to be classified as male or female So we then added other ( O ) We had no choice this was specifically written for the state and they were the major user of the program…oh well, nothing is black or white is it?

Wednesday, May 24, 2006

Spatial Indexes In SQL Server 2005

Maybe you work with Oracle as well as SQL server 2005 and you use Oracle Spatial and Oracle Locator. Well there is a way to implement this in SQL Server 2005.
Read the article "Using Table Valued Functions in SQL Server 2005 to Implement a Spatial Data Library" on MSDN.

This article explains how to add spatial search functions (point-near-point and point in polygon) to Microsoft SQL Server 2005 using C# and table-valued functions. It is possible to use this library to add spatial search to your application without writing any special code. The library implements the public-domain C# Hierarchical Triangular Mesh (HTM) algorithms from Johns Hopkins University. That C# library is connected to SQL Server 2005 using a set of scalar-valued and table-valued functions. These functions act as a spatial index.

SQL Server Pics From Version 1.0 On

Euan Garden has posted pictures of the SQL server product boxes since version 1.0
Did you know version 1.0 came out on Floppy's for OS/2?
Check out the pictures here
http://blogs.msdn.com/euanga/archive/2006/05/23/597677.aspx

Tuesday, May 23, 2006

SQL Server Teaser: Isolation Level

In SQL server 2000 there are 4 isolation levels

Read uncommitted which can produce a dirty read, a repeatable read and a phantom read
Read committed which can produce a repeatable read and a phantom read
Repeatable read which can produce a phantom read
Serializable

What if I tell you that there is a 5th one, no it’s not snapshot isolation we are dealing with SQL Server 2000 here

The question is how do you set it and what is the (appropriate) name of this isolation level
I will post the answer on Memorial Day if no nobody answers by then

Monday, May 22, 2006

Cumulative Hotfix Package (build 2153) For SQL Server 2005 Available

So just after sp1 we already have a hotfix
The following has been fixed

SQL Server 2005 Post-SP1 hotfixes that are included in SQL Server 2005 build 9.0.2153
SQL Bug number Description
410 Dimension security does not support visual totals on a parent attribute that is in a parent-child dimension.
433 If you execute an ALTER DDL statement that removes dimension hierarchies from the database without removing the hierarchies from the cubes, an access violation may occur.
447 Processing performance on multiprocessor computers is somewhat slower than expected.
459 Under very specific and rare circumstances, a deadlock can occur between two or more sessions in SQL Server 2005. In this case, the internal deadlock monitoring mechanism does not detect the deadlock between the sessions.
461 Page breaks may not be respected when a table is rendered in Microsoft Office Excel if that report contains tables with the conditions of NoRows=true and NoRowsMessage=null, and the table contains a table header or footer.
464 The leaf page of a non-clustered index may have only one index row.
491 On a SQL Server 2005 installation that uses Turkish collation, you cannot view job steps that are of SSIS type from SQL Server Management Studio. If a job has SSIS job steps only, you receive an "Index was outside the bounds of the array. (SqlManagerUI)" error message when you click the Steps tab.
499 A Multidimensional Expressions (MDX) query that involves non-trivial custom rollups when it runs complex sum-like aggregate expressions runs much slower on SQL Server 2005 than on SQL Server 2000.
531 When you run a query or a stored procedure in SQL Server 2005, certain conditions can cause the SQL Server optimizer not to be able to produce an execution plan.
541 An ADOMD.NET application that runs through an IXMLA provider is limited to 32 maximum user sessions.
554 When you query the IS_PLACEHOLDERMEMBER member property for a dimension member by using the .Properties("IS_PLACEHOLDERMEMBER") function in MDX, the value that is returned is always FALSE. This behavior occurs even for members that are placeholders in the dimension.
584 When a report contains an image that is external or is a resource in Reporting Services and the report uses the interactive sorting functionality, the image is replaced by a red x.
606 The MDSCHEMA_PROPERTIES row set returns the wrong data type for some user-defined properties.
608 If you include a subreport in a group footer and you enable the HideDuplicates property in a detail row on a grouping item, SQL Server 2005 Reporting Services raises an internal error when you try to export the report. The error also occurs when you click Print Preview on the Preview tab in Report Designer.
614 SQL Server 2005 Reporting Services may intermittently display a System.NullReferenceException exception in the ReportSnapshot.EnsureAllStreamsAreClosed procedure. This behavior generates a mini-dump.
615 The SQL Server 2005 Analysis Services service does not start when you use Japanese_Unicode_CI_AS as a collation for the instance of SQL Server 2005 Analysis Services.
617 When you try to run a linked report that uses the User!UserID variable and the execution properties are set to use Snapshots, you receive the following error message:
3wp!library!1!1/27/2006-13:10:27:: e ERROR: Throwing Microsoft.ReportingServices.Diagnostics.Utilities.InternalCatalogException:
627 In the Japanese version of SQL Server 2005, the fulltext schedule creation in SQL Server Management Studio fails, and you receive the following error message:
Specified @category_name ('フルテキスト'localized string) not found (Microsoft SQL Server, Error: 14262)
636 A memory leak may occur in the common language runtime (CLR) if you pass a CLR user-defined data type as a parameter by using a stored procedure or the sp_executesql procedure.
652 If Y-axis margins are enabled and the minimum Y-value of all data points is a "nice" integer value, the Y-axis minimum value will be adjusted to a lower value. For example, the chart control rounds 4.0 to 4 to create a "nice" integer value. However, the generated Y-axis labels are incorrect.
664 You run SQL Server 2005 Analysis Services builds 9.00.1520 through 9.00.1539 or SQL Server 2005 SP1. In this case, an MDX query that includes a distinct count measure may not return results. This behavior only occurs if the measure group that is queried contains multiple partitions.
708 The SQL Server Integration Services (SSIS) service does not start when you install SQL Server 2005 SP1
752 The MSMQTask.dll and dependencies are deleted when the fix that is described in the Microsoft Knowledge Base article 910070 is applied.
40000100 If the log reader agent and the DBCC DBREINDEX or ALTER INDEX ...REBUILD commands run at the same time, the log reader agent may miss some transactions to the distributor.
40000102 When you configure Log Shipping and add two or more secondary servers to Log Shipping, only the secondary server that you added last is monitored by the monitor server and the other secondary servers are not.
40000108 The SQL Server 2005 Database Engine Tuning Advisor may unexpectedly exit if indexed views are part of the workload that is tuned.
40000110 The YTD calculation in cubes that use parent-child dimensions may be incorrect.
50000104 The SqlCommandBuilder.DeriveParameters(SqlCommand) method returns an exception when you execute a stored procedure with an input parameter that is a typed XML type.
50000112 A member of the SQL Server 2005 Analysis Services database administrator role cannot add or delete perspectives. Only a member of the SQL Server 2005 Analysis Services server role can perform such tasks.
50000115 A Report Server Model may use an Analysis Services data source. MDX queries are generated from Report Builder based on the Report Server Model. Dates in the MDX queries have the month and day parts transposed.
50000116 A FETCH operation that uses a KEYSET cursor may be slow if the target table for the cursor has a clustered index and non-clustered indexes. This behavior occurs if the clustered index is the best choice for the FETCH operation and if the SELECT query for the cursor is very simple.
50000120 In SQL Server 2005 Analysis Services, calculated members that are defined in the cube MDX script are resolved before any session or query defined calculations. The only exceptions are certain scenarios with the Aggregate and VisualTotals functions. This behavior is a change from the SQL Server 2000 Analysis Services behavior. In SQL Server 2000 Analysis Services, the Solve Order could be used explicitly to insert a session or a query-defined calculation between two cube level calculations. This change in behavior may cause the query or session scope calculated members not to return the results that you want in SQL Server 2005 Analysis Services. Cumulative hotfix package for SQL Server 2005 (build 9.0.2153) introduces new syntax that allows for control over the scope in which a calculated member is evaluated.
50000151 When you run a SELECT query to return any duplicate rows in a table, incorrect results are returned if the query plan contains a hash aggregation and if one or more of the columns in the table contain a combination of empty strings and undefined characters.
50000154 You may receive a variablelock list error message when you run multiple instances of a package.

You can download the hotfix here

Friday, May 19, 2006

Customers who bought this item also bought...

Customers who bought this item also bought....by now you probably know that I am talking about Amazon.com
So let's try to implement our own (simple) Amazon system
We will call it People who bought this song also bought these songs
So basically we will create 3 tables (ArtistTable, Songtable and SongPaid)
These are very basic tables and they are missing a lot of info, I created them so that the query is nice and simple
So if person A buys song 3 then we want to return all the songs (except song 3) from customers who also bought song 3 (except customer A)
we we also list how many times a song was bought, so that we can see what the most popular songs were that people bought who also bought the same song that we bought
Let's create our tables first

CREATE TABLE ArtistTable (ID INT PRIMARY KEY, ArtistName VARCHAR(500))
INSERT INTO ArtistTable VALUES(1,'Red Hot Chilli Peppers')
INSERT INTO ArtistTable VALUES(2,'Pearl Jam')
INSERT INTO ArtistTable VALUES(3,'Astral Projection')
INSERT INTO ArtistTable VALUES(4,'Chemical Brothers')

CREATE TABLE Songtable (ID INT PRIMARY KEY , SongName VARCHAR(500),ArtistID INT)
INSERT INTO Songtable VALUES(1,'Under The Bridge',1)
INSERT INTO Songtable VALUES(2,'Black',2)
INSERT INTO Songtable VALUES(3,'Mahadeva',3)
INSERT INTO Songtable VALUES(4,'Block Rockin Beats',4)
INSERT INTO Songtable VALUES(5,'Setting Sun',4)
INSERT INTO Songtable VALUES(6,'Alive',1)
INSERT INTO Songtable VALUES(7,'Give It Away Now',2)

CREATE TABLE SongPaid (SongID INT, CustID INT)
INSERT INTO SongPaid VALUES(1, 1)
INSERT INTO SongPaid VALUES(2, 1)
INSERT INTO SongPaid VALUES(1, 2)
INSERT INTO SongPaid VALUES(3, 3) -- Current Buyer (3)
INSERT INTO SongPaid VALUES(3, 4) -- Another Buyer (4)
INSERT INTO SongPaid VALUES(4, 4) -- Also bought by 4 ('Block Rockin Beats')
INSERT INTO SongPaid VALUES(5, 4) -- Also bought by 4 ('Setting Sun')
INSERT INTO SongPaid VALUES(4, 5)
INSERT INTO SongPaid VALUES(4, 6)
INSERT INTO SongPaid VALUES(5, 6)
INSERT INTO SongPaid VALUES(3, 7) -- Another Buyer (4)
INSERT INTO SongPaid VALUES(7, 7) -- Also bought by 7 ('Give It Away Now')
INSERT INTO SongPaid VALUES(4, 7) -- Also bought by 7 ('Block Rockin Beats')

CREATE INDEX ix_SongPaid ON SongPaid(SongID,CustID)

/*Now let's do the query with person 3 and song 3
our result set will be

SongID,SongName,ArtisName,SongCount
4,Block Rockin Beats,Chemical Brothers,2
5,Setting Sun Chemical Brothers,1
7,Give It Away Now,Pearl Jam,1

*/


DECLARE @SongId INT, @CustId INT
SELECT @SongId = 3,@CustId =3


SELECT s1.SongID ,st.Songname,a.ArtistName, Count(*) AS SongCount
FROM SongPaid s1
JOIN Songtable st on s1.SongID = st.ID
JOIN ArtistTable a on st.ArtistID = a.ID
WHERE EXISTS (
SELECT custid FROM SongPaid
WHERE SongID =@SongId AND CustID =s1.CustID )
AND SongID <> @SongId --Don't return the song we bought
AND CustID <> @CustId --Skip other songs we bought by skipping our customerid
GROUP BY s1.SongID ,st.Songname,a.ArtistName


The songs and bands I used in the tables are real so in case you don't know them here is some additional info with an Amazon link to the CD
'Under The Bridge' and 'Give It Away Now' are from the CD Blood Sugar Sex Magik by the Red Hot Chilli Peppers
'Black' and 'Alive' are from the CD Ten by Pearl Jam
'Mahadeva' is from the CD In The Mix by Astral Projection
'Setting Sun' and 'Block Rockin Beats'is from the CD Singles 93-03 by the Chemical Brothers

Wednesday, May 17, 2006

List All Tables Without An Index

Here are 2 quick ways to list all tables in your database without an index
The first query is using sysobjects and the second query is using the INFORMATION_SCHEMA.TABLES view

SELECT name AS TableName
FROM sysobjects s
WHERE type = 'U'
AND OBJECTPROPERTY(id , 'TableHasIndex' ) = 0
ORDER BY TableName


SELECT TABLE_NAME AS TableName
FROM INFORMATION_SCHEMA.TABLES
WHERE TABLE_TYPE ='BASE TABLE'
AND OBJECTPROPERTY(OBJECT_ID(TABLE_NAME) , 'TableHasIndex' ) = 0
ORDER BY TableName

Tuesday, May 16, 2006

SQL Compare Without The Price Tag

You want to know the differences between 2 tables but don’t have the money to pay for SQL Compare by red-gate software and your trial version has expired ;-(
Let me say first of all that SQL Compare is probably the best third party product around for SQL Server (along with SQL LITESPEED by QUEST Software)

In order to see what rows are in table1 and not in table 2 and vice versa you can do 2 left joins, 2 right joins or 1 left and 1 right join. To get the rows that are different you can use CHECKSUM
Let’s get started…

--let's copy over 20 rows to a table named authors2
SELECT TOP 20 * INTO tempdb..authors2
FROM pubs..authors

--update 5 records by appending X to the au_fname
SET ROWCOUNT 5


UPDATE tempdb..authors2
SET au_fname =au_fname +'X'


--Set rowcount back to 0
SET ROWCOUNT 0

--let's insert a row that doesn't exist in pubs
INSERT INTO tempdb..authors2
SELECT '666-66-6666', au_lname, au_fname, phone, address, city, state, zip, contract
FROM tempdb..authors2
WHERE au_id ='172-32-1176'

--*** The BIG SELECT QUERY --***

--Not in Pubs
SELECT 'Does Not Exist On Production',t2.au_id
FROM pubs..authors t1
RIGHT JOIN tempdb..authors2 t2 ON t1.au_id =t2.au_id
WHERE t1.au_id IS NULL
UNION ALL
--Not in Temp
SELECT 'Does Not Exist In Staging',t1.au_id
FROM pubs..authors t1
LEFT JOIN tempdb..authors2 t2 ON t1.au_id =t2.au_id
WHERE t2.au_id IS NULL
UNION ALL
--Data Mismatch
SELECT 'Data Mismatch', t1.au_id
FROM( SELECT BINARY_CHECKSUM(*) AS CheckSum1 ,au_id FROM pubs..authors) t1
JOIN(SELECT BINARY_CHECKSUM(*) AS CheckSum2,au_id FROM tempdb..authors2) t2 ON t1.au_id =t2.au_id
WHERE CheckSum1 <> CheckSum2

--Clean up
DROP TABLE tempdb..authors2
GO

You can run this whole script in 1 shot

Let me finish by saying that this code should just be used for quick checking. SQL Compare is far superior to this and will also generate the scripts for you to synchronize the 2 tables. So if you only need to do something like this once then use this code or download the SQL compare trial version. If you do this on a regular basis then you should seriously consider getting SQL compare

I am not being paid or endorsed in any way by red-gate, I just happen to like their product a lot and it saved me a lot of time and trouble once a developer went on vacation

[Edit]
SQL server MVP Louis Davidson has posted a comment and this was his approach
Basically a FULL OUTER JOIN instead of a UNION
So here is his query


SELECT CASE WHEN t1.au_id IS NULL
AND t2.au_id IS NOT NULL
THEN 'Does Not Exist On Production'
WHEN t1.au_id IS NOT NULL
AND t2.au_id IS NULL
THEN 'Does Not Exist In Staging'
ELSE 'Data Mismatch' END,
COALESCE(t1.au_id, t2.au_id) AS au_id
FROM (SELECT *, BINARY_CHECKSUM(*) AS bc FROM pubs..authors) AS t1
FULL OUTER JOIN (SELECT *, BINARY_CHECKSUM(*) AS bc FROM tempdb..authors2) AS t2
ON t1.au_id =t2.au_id
WHERE t1.au_id IS NULL
OR t2.au_id IS NULL
OR t1.bc <> t2.bc

Louis is also the author of the just released Pro SQL Server 2005 Database Design and Optimization book

[/edit]

Monday, May 15, 2006

Convert Millisecond To "hh:mm:ss" Format

This question was posted in the Transact-SQL MSDN Forum
You have a column in your table that stores duration in milliseconds
You would like to output this in the following format ‘hh:mm:ss’
So for example 11354013 would be 03:09:14:013 and 86399998 milliseconds would be 23:59:59:997 But be careful because 86399998 will return 23:59:59:997 but 86399999 will return 00:00:00:000 (it will round up to 1 day, 0 hours etc etc)

So let’s see how to do that. Basically all you have to do is add the milliseconds to a zero date and then convert that to varchar with a style of 114 (hh:mi:ss:mmm(24h))

DECLARE @SomeMilliSecondsNumber INT
SELECT @SomeMilliSecondsNumber =11354013

SELECT CONVERT(VARCHAR,DATEADD(ms,@SomeMilliSecondsNumber,0),114)
--03:09:14:013
GO


DECLARE @SomeMilliSecondsNumber INT
SELECT @SomeMilliSecondsNumber =86399998

SELECT CONVERT(VARCHAR,DATEADD(ms,@SomeMilliSecondsNumber,0),114)
--23:59:59:997
GO

DECLARE @SomeMilliSecondsNumber INT
SELECT @SomeMilliSecondsNumber =86399999

SELECT CONVERT(VARCHAR,DATEADD(ms,@SomeMilliSecondsNumber,0),114)
--00:00:00:000
GO

Friday, May 12, 2006

Syntax Error Converting The Varchar Value To A Column Of Data Type Int

There was a question today on the microsoft.public.sqlserver.programming forum where a person had a problem with an in predicate
The error that was raised was 'Syntax error converting the varchar value '33333A' to a column of data type int'
The error is a little bit misleading
Let's say you have the following query

SELECT * FROM SomeTable WHERE SomeColumn in (12345,22222)

And you get that error, you might think I am not converting anything why am I getting this error?
If the query was the one below then it would make sense

SELECT CONVERT(INT,SomeColumn) FROM SomeTable...

Now let's see what causes this error


--Create the test table and insert 3 rows of data
CREATE TABLE SomeTable (id INT,SomeColumn VARCHAR(49))
INSERT INTO SomeTable VALUES(1,'12345')
INSERT INTO SomeTable VALUES(1,'22222')
INSERT INTO SomeTable VALUES(1,'33333')

--let's select from that table
--no problem
SELECT * FROM SomeTable WHERE SomeColumn in (12345,22222)


--Let's insert something with a non numeric value
INSERT INTO SomeTable VALUES(1,'33333A')


--Let's try again
SELECT * FROM SomeTable WHERE SomeColumn in (12345,22222)
--Oops error
--Server: Msg 245, Level 16, State 1, Line 1
--Syntax error converting the varchar value '33333A' to a column of data type int.



--The trick is to put quotes around the values
--And now it works
SELECT * FROM SomeTable WHERE SomeColumn in ('12345','22222')

--Clean up
DROP TABLE SomeTable

Changing The Schema Of A Table In SQL Server 2005

Since I am still in unpacking and putting up light-fixtures mode I decided to link to a cool post on the SQL Doctor's blog
It covers the process of changing the schema for a table
All you need to do is this
ALTER SCHEMA [schemaName] TRANSFER [otherSchemaName].[objectName]

Here is the link (http://spaces.msn.com/drsql/blog/cns!80677FB08B3162E4!1041.entry)

And don't let Fred, Barney, Wilma or Betty fool you

Thursday, May 11, 2006

String Or Binary Data Would Be Truncated

I have seen a couple of Google searches hitting this blog with the search 'String or binary data would be truncated'
Basically what the error message 'String or binary data would be truncated' means is that the data that you are trying to fit into a column inside a table won't fit because the column isn't wide enough
A lot of times this occurs when you are inserting data into 1 table from another table

For example you have a table on some database server and you will need to import that data into a new table
You will have to store Unicode in the future so you make one of the columns nvarchar, you don't bother to check for the maximum length of the data in the original table and just make you column nvarchar(4000) (the max in SQL server 2000)
Now let's test a couple of these cases

--Create the 'original' table
CREATE TABLE TestData (ID int,
SomeValue VARCHAR(5000),
SomeOtherValue VARCHAR(6))


--add some data
INSERT INTO TestData VALUES (1,REPLICATE('A',4002),'abcdef')
INSERT INTO TestData VALUES (2,'123','abcde')
INSERT INTO TestData VALUES (3,'123','abcdef')


--Create the 'new' table
CREATE TABLE TestDataTruncate2 (ID INT,
SomeValue NVARCHAR(4000),
SomeOtherValue VARCHAR(5)) --Oops 1 less than in the original table

--Fails
INSERT INTO TestDataTruncate
SELECT * FROM TestData
WHERE ID =1
/*
Server: Msg 8152, Level 16, State 9, Line 1
String or binary data would be truncated.
The statement has been terminated.

Fails because nvarchar can only hold 4000 characters
*/



--No Problem
INSERT INTO TestDataTruncate
SELECT * FROM TestData
WHERE ID =2

--Fails
INSERT INTO TestDataTruncate
SELECT * FROM TestData
WHERE ID =3
/*
Server: Msg 8152, Level 16, State 9, Line 1
String or binary data would be truncated.
The statement has been terminated.

Fails because SomeOtherValue has 6 charaters of data
but the column is only 5 characters in the new table
*/



--Instead of opening up 2 windows and comparing the tables column by column
--you can use the query below to return all that info in a resultset

SELECT TABLE_NAME,COLUMN_NAME,DATA_TYPE,CHARACTER_MAXIMUM_LENGTH,ORDINAL_POSITION
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME ='TestDataTruncate'
UNION ALL
SELECT TABLE_NAME,COLUMN_NAME,DATA_TYPE,CHARACTER_MAXIMUM_LENGTH,ORDINAL_POSITION FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME ='TestData'
ORDER BY ORDINAL_POSITION,TABLE_NAME

--you can also create a self join and return only the columns
--where the name is the same but the column size is different

SELECT t1.COLUMN_NAME,t1.DATA_TYPE,
t1.CHARACTER_MAXIMUM_LENGTH,t2.CHARACTER_MAXIMUM_LENGTH,t1.ORDINAL_POSITION FROM INFORMATION_SCHEMA.COLUMNS t1
JOIN INFORMATION_SCHEMA.COLUMNS t2 ON t1.COLUMN_NAME =t2.COLUMN_NAME AND t1.TABLE_NAME <> t2.TABLE_NAME
WHERE t1.TABLE_NAME ='TestDataTruncate'
AND t2.TABLE_NAME ='TestData'
AND t1.CHARACTER_MAXIMUM_LENGTH <> t2.CHARACTER_MAXIMUM_LENGTH


--Strangely enough when you try to assign values to a variable
--it will truncate it without an error message
DECLARE @chvCity VARCHAR(8)
SELECT @chvCity ='Princeton'
SELECT @chvCity


--Clean up this mess ;-)
DROP TABLE TestDataTruncate,TestData

Tuesday, May 09, 2006

Moved Into A New House

I have moved into a new house over the last week. If you are thinking of selling your house and buying a new one at the same time then either wait until you hit the jackpot/win the lotto or wait until your old house is paid off. This has been the most stressful experience of my life. The buyer of my house was giving me a hard time, then she couldn’t get a mortgage commitment and after that she decided to switch mortgage companies. Nothing was certain until the last millisecond. And the fact that my wife is 7 months pregnant with twins didn’t help either with this ordeal. I won’t have an internet connection until this weekend so I probably won’t post a lot this week
This post and the one that I will do after the twins are born are probably the only two non SQL related posts that will ever be on this blog, hopefully you don’t mind

Tuesday, May 02, 2006

Top SQL Server Google Searches For April 2006

These are the top SQL Searches on this site for the month of April. I have left out searches that have nothing to do with SQL Server or programming. Here are the results...

sql server CONNECTIVITY ERRORS thro dsl after installing xp sp2
stored procedure divide
execution plans
String or binary data would be truncated
rank() performance
copy table to other server
rank() performance sql
SQL SERVER 2005 SELECT FIELD AS VARIABLE
ToBase64String sql procedure
Server: Msg 7357, Level 16, State 2, Line 1
"= isNull"

It's always interesting to see what people are searching for, I left the money laundry one in this list because I thought it was kind of amusing
I always like to look at this list so that I know what people are interested in and I can write a little thing about it

Top 10 Articles Of All Time

Below are the top 10 articles of all time according to Google Analytics (updated 6/1/2006)
There is a link to it right below the Google Search box, I will update this once a month

1 SQL Query Optimizations
2 Five Ways To Return Values From Stored Procedures
3 Login failed for user 'sa'. Reason: Not associated with a trusted SQL Server connection. SQL 2005
4 COALESCE And ISNULL Differences
5 SQL Server 2005 Free E-Learning
6 OPENROWSET And Excel Problems
7 Do You Know How Between Works With Dates?
8 NULL Trouble In SQL Server Land
9 Find all Primary and Foreign Keys In A Database
10 Fun With SQL Server Update Triggers

Top 5 SQL Server Posts For April 2006

Below are the top 5 posts according to Google Analytics for the month of April

Login failed for user 'sa'. Reason: Not associated with a trusted SQL Server connection. SQL 2005
OPENROWSET And Excel Problems
SQL Query Optimizations
SQL Server 2005 Free E-Learning
Feature Pack for Microsoft SQL Server 2005

A Developer’s Guide to SQL Server 2005


The follow up to the highly succesful book 'A First Look at SQL Server 2005 for Developers' has been published by Addison-Wesley
The book is 1008 pages long

From the site:

Few technologies have been as eagerly anticipated as Microsoft SQL Server 2005. Now, two SQL Server insiders deliver the definitive hands-on guide--accurate, comprehensive, and packed with examples. A Developer's Guide to SQL Server 2005 starts where Microsoft's documentation, white papers, and Web articles leave off, showing developers how to take full advantage of SQL Server 2005's key innovations. It draws on exceptional cooperation from Microsoft's SQL Server developers and the authors' extensive access to SQL Server 2005 since its earliest alpha releases.

You'll find practical explanations of the new SQL Server 2005 data model, built-in .NET hosting, improved programmability, SQL:1999 compliance, and much more. Virtually every key concept is illuminated via sample code that has been fully updated for and tested with the shipping version of the product.

Key coverage includes
Using SQL Server 2005 as a .NET runtime host: extending the server while enhancing security, reliability, and performance
Writing procedures, functions, triggers, and types in .NET languages
Exploiting enhancements to T-SQL for robust error-handling, efficient queries, and improved syntax
Effectively using the XML data type and XML queries
Implementing native SQL Server 2005 Web Services
Writing efficient, robust clients for SQL Server 2005 using ADO.NET, classic ADO, and other APIs
Taking full advantage of user-defined types (UDTs), query notifications, promotable transactions, and multiple active result sets (MARS)
Using SQL Management Objects (SMO), SQL Service Broker, and SQL Server Notification Services to build integrated applications


Download the Sample Chapter related to this title.

The Amazon link is here

Monday, May 01, 2006

Check For Not In Table Values (16 Different Ways And Counting)

How many times did you see a question like this:
how do I select all rows from a table where the column value is not 'ValueA' and not 'ValueB'
Well I decided to figure out how many different ways there are to do such a query
I came up with 16 different ways so far, some of them will cause a table scan, some of them will cause a index seek instead
You can test them out for yourself to see which ones perform the best

CREATE TABLE #TestTable (
Objects VARCHAR(20) PRIMARY KEY)

INSERT INTO #TestTable
VALUES ('boat')

INSERT INTO #TestTable
VALUES('ship')

INSERT INTO #TestTable
VALUES('car')

INSERT INTO #TestTable
VALUES('bus')

INSERT INTO #TestTable
VALUES ('airplane')


-- #1 <>
SELECT *,1
FROM #TestTable
WHERE Objects <> 'boat'
AND Objects <> 'ship'

-- #2 !=
SELECT *,2
FROM #TestTable
WHERE Objects != 'boat'
AND Objects != 'ship'


-- #3 NOT
SELECT *,3
FROM #TestTable
WHERE NOT (Objects = 'boat'
OR Objects = 'ship')


-- #4 NOT IN
SELECT *,4
FROM #TestTable
WHERE Objects NOT IN ('boat',
'ship')


-- #5 ALL
SELECT *,5
FROM #TestTable
WHERE Objects <> ALL (SELECT 'boat'
UNION ALL
SELECT 'ship')


-- #6 ANY
SELECT *,6
FROM #TestTable
WHERE NOT Objects = ANY (SELECT 'boat'
UNION ALL
SELECT 'ship')


-- #7 SOME
SELECT *,7
FROM #TestTable
WHERE NOT Objects = SOME (SELECT 'boat'
UNION ALL
SELECT 'ship')


-- #8 NOT IN with subquery
SELECT *,8
FROM #TestTable
WHERE Objects NOT IN (SELECT 'boat'
UNION ALL
SELECT 'ship')


-- #9 NOT EXISTS
SELECT *,9
FROM #TestTable T
WHERE NOT EXISTS (SELECT *
FROM (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
WHERE F.Objects = T.Objects)

-- #10 LEFT OUTER JOIN
SELECT T.*,10
FROM #TestTable T
LEFT OUTER JOIN (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
ON T.Objects = F.Objects
WHERE F.Objects IS NULL


-- #11 CHARINDEX
SELECT *,11
FROM #TestTable
WHERE CHARINDEX('-' + Objects + '-','-boat-ship-') = 0


-- #12 RIGHT OUTER JOIN
SELECT T.*,12
FROM (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
RIGHT OUTER JOIN #TestTable T
ON T.Objects = F.Objects
WHERE F.Objects IS NULL
AND T.Objects IS NOT NULL


-- #13 FULL OUTER JOIN
SELECT T.*,13
FROM (SELECT 'boat' AS Objects
UNION ALL
SELECT 'ship') F
FULL OUTER JOIN #TestTable T
ON T.Objects = F.Objects
WHERE F.Objects IS NULL


-- #14 PATINDEX
SELECT *,14
FROM #TestTable
WHERE PATINDEX('%-%' + Objects + '%-%','-boat-ship-') = 0


-- #15 PARSENAME
SELECT *,15
FROM #TestTable
WHERE Objects <> PARSENAME('boat.ship',1)
AND Objects <> PARSENAME('boat.ship',2)


-- #16 REVERSE and NOT IN
SELECT *,16
FROM #TestTable
WHERE REVERSE(Objects) NOT IN ('taob','pihs')

-- And here is another one that BugsBunny suggested in a comment
-- #17 NOT LIKE

SELECT *,17
FROM #TestTable
WHERE ',' + 'boat,ship' + ',' NOT LIKE '%,' + Objects + ',%'


DROP TABLE #TestTable

So that's it, if you know of another way, leave me a comment or email me and I will add it to this list

Friday, April 28, 2006

Splitting City, State and Zipcode From One Column (Or Flat file)

Sometimes you deal with vendors, customers or agencies and you have to do a file exchange
Of course the format of these files is always dictated by these people so what do you do when a CityStateZip column contains values like “Long Island City, NY 10013” or like “Princeton,NJ 08536 “? But in your Database it is normalized of course, and you have 3 columns. You will have to use a combination of LEFT, LTRIM, SUBSTRING, REPLACE and RIGHT. I am taking into account that there could be spaces in the column or even spaces in the name (New York City) So let’s get started and see what we get

CREATE TABLE #TestCityStateZip (csz CHAR(49))
INSERT INTO #TestCityStateZip VALUES ('city ,st 12223')
INSERT INTO #TestCityStateZip VALUES ('New York City,NY 10028')
INSERT INTO #TestCityStateZip VALUES ('Princeton , NJ 08536')
INSERT INTO #TestCityStateZip VALUES ('Princeton,NJ 08536 ')
INSERT INTO #TestCityStateZip VALUES ('Long Island City, NY 10013')
INSERT INTO #TestCityStateZip VALUES ('Long Island City, NY 10013 ')
INSERT INTO #TestCityStateZip VALUES ('Long Island City , NY 10013')
INSERT INTO #TestCityStateZip VALUES ('Long Island City ,NY 10013 ')


SELECT LEFT(csz,CHARINDEX(',',csz)-1)AS City,
LEFT(LTRIM(SUBSTRING(csz,(CHARINDEX(',',csz)+1),4)),2) AS State,
RIGHT(RTRIM(csz),CHARINDEX(' ',REVERSE(RTRIM(csz)))-1) AS Zip
FROM #TestCityStateZip

GO

Tuesday, April 25, 2006

SQL Server 2005 SP1 Books Online (April 2006) For Download

SQL Server 2005 SP1 Books Online (April 2006) is available for download
Download an updated version of Books Online for Microsoft SQL Server 2005. Books Online is the primary documentation for SQL Server 2005. The April 2006 update to Books Online contains new material and fixes to documentation problems reported by customers after SQL Server 2005 was released. Refer to "New and Updated Books Online Topics" for a list of topics that are new or updated in this version. Topics with significant updates have a Change History table at the bottom of the topic that summarizes the changes. Beginning with the April 2006 update, SQL Server 2005 Books Online reflects product upgrades included in SQL Server 2005 Service Pack 1 (SP1).

The download is 121 MB

http://www.microsoft.com/downloads/details.aspx?FamilyID=be6a2c5d-00df-4220-b133-29c1e0b6585f&DisplayLang=en

Monday, April 24, 2006

How To Script Multiple Jobs Or Stored Procedures In SQL Server 2005 Management Studio

One of the biggest complaints of people who moved from SQL server 2000 to SQL Server 2005 is the inability to script multiple objects. Well that feature is still available only it’s kind of hidden
Click the Jobs or Stored Procedures folder in SQL Server Management Studio, and then hit the F7 key; this will bring up the Summary pane. Highlight all the Jobs or Stored Procedures that you want to script using a combination of Shift and Ctrl keys, then right click, Script Job/Stored Procedure as..., and then choose where to save this script to. An image of how to script Stored Procedures is below

Thursday, April 20, 2006

Wednesday, April 19, 2006

Microsoft Releases SQL Server 2005 Service Pack 1

Microsoft Corp. today announced the availability of Microsoft® SQL Server™ 2005 Service Pack 1 (SP1), the product’s first major milestone since the launch of SQL Server 2005 only a few months ago. The release follows the remarkable reception, momentum and feedback offered by customers and partners in 92 countries. This has informed and accelerated Microsoft’s multiyear vision of Your Data, Any Place, Any Time. Microsoft SQL Server 2005 SP1 includes production-ready Database Mirroring functionality and the new SQL Server Management Studio Express, along with additional updates to SQL Server 2005 Express Edition targeted at users such as independent software vendors (ISVs) wanting to take advantage of greater functionality.

SP1 delivers production-ready Database Mirroring functionality for continuous availability. This complements the existing Always On Technologies in SQL Server 2005 such as failover clustering, database snapshots, snapshot isolation and log shipping. Since launch, Database Mirroring has been extensively tested by Microsoft and its thriving customer community to help ensure that it provides the high availability necessary for the largest customer deployments. Today more than 20 SQL Server customers have deployed Database Mirroring into production

SP1 extends business insight to smaller-scale customers and ISVs using the free SQL Server 2005 Express Edition, which includes SQL Server Reporting Services (SSRS) functionality, Full Text Search and the newly released SQL Server Management Studio Express. This optional set of capabilities is already receiving positive feedback from Microsoft’s ISV partners

SP1 also advances dynamic applications with the new SQL Server Management Studio Express, a simplified graphical management environment for SQL Server Express Edition. This tool builds on Microsoft’s commitment to extending the breadth of the SQL Server family to support all the data storage needs of Microsoft’s customers. It also complements the recently announced SQL Server Everywhere Edition, a lightweight and rich subset of capabilities found in other SQL Server editions, targeted for application embedded storage on clients. SQL Server Everywhere Edition is targeted to be available as a Community Technology Preview (CTP) this summer and released by year end.

For end-to-end business insight, SP1 extends SSRS to support enterprise reporting on SAP NetWeaver Business Intelligence with two new components in SP1: a Microsoft .NET Data provider for SAP NetWeaver Business Intelligence and a new MDX Query Designer. This new functionality extends the benefits of enterprise reporting with SQL Server by enabling SAP customers to easily create and manage reports on information inside any SAP BW data warehouse

Inside Microsoft SQL Server 2005: T-SQL Querying

How did I miss this? SQL Server MVP Itzik Ben-Gan has published his latest book: Inside Microsoft SQL Server 2005: T-SQL Querying. For all of you who read SQL Server magazine you probably know Itzik from his great SQL tips and puzzles articles.
Take a detailed look at the internal architecture of T-SQL—and unveil the power of set-based querying—with comprehensive reference and advice from the experts. Database developers and administrators get best practices, sample databases, and code to master the intricacies of the programming language—solving complex problems with real-world solutions.




Discover how to:

•Understand logical and physical query processing
•Apply a methodology to optimize query tuning
•Solve relational division problems
•Use CTEs and ranking functions to simplify and optimize solutions
•Aggregate data with various techniques, including tiebreakers, pivoting, histograms, and grouping factors
•Use the TOP option in a query to modify data
•Query specialized data structures with recursive logic, materialized path, or nested sets solutions
•PLUS—Improve your logic and get to the heart of querying problems with logic puzzles


The book is 640 printed pages, below is the table of contents

Chapter 01 - Logical Query Processing
Chapter 02 - Physical Query Processing
Chapter 03 - Query Tuning
Chapter 04 - Subqueries, Table Expressions and Ranking Functions
Chapter 05 - Joins and Set Operations
Chapter 06 - Aggregating and Pivoting Data
Chapter 07 - TOP and APPLY
Chapter 08 - Data Modification
Chapter 09 - Graphs, Trees, Hierarchies and Recursive Queries
Appendix A - Logic Puzzles

I could not find a sample chapter yet but as soon one is available I will create a link to it

If you are intereseted in purchasing this book the Amazon link is here

Tuesday, April 18, 2006

Use PARSENAME, CHARINDEX, PATINDEX or SUBSTRING To Grab Values Up To A Certain Character

This is a question that came up yesterday in the Getting started with SQL Server MSDN forum (http://forums.microsoft.com/MSDN/ShowPost.aspx?PostID=353250&SiteID=1)
A person wanted to use a MID function in SQL Server, There is no MID function in SQL Server but there are at least 4 ways to implement what the person tried to accomplish
Basically if the data looked like this

aaa-bbbbb
ppppp-bbbbb
zzzz-xxxxx

Then we need to grab everything up until the minus sign
So let's get started

CREATE TABLE #Test (
SomeField VARCHAR(49))

INSERT INTO #Test
VALUES ('aaa-bbbbb')

INSERT INTO #Test
VALUES ('ppppp-bbbbb')

INSERT INTO #Test
VALUES ('zzzz-xxxxx')

--using PARSENAME
SELECT PARSENAME(REPLACE(SomeField,'-','.'),2)
FROM #Test

--using LEFT and CHARINDEX
SELECT LEFT(SomeField,CHARINDEX('-',SomeField) - 1)
FROM #Test

--using LEFT and PATINDEX
SELECT LEFT(SomeField,PATINDEX('%-%',SomeField) - 1)
FROM #Test


--using CASE SUBSTRING and LEFT
SELECT CASE SUBSTRING(SomeField,4,1)
WHEN '-' THEN LEFT(SomeField,3)
ELSE LEFT(SomeField,4)
END
FROM #Test


--clean up
DROP TABLE #Test

Monday, April 17, 2006

Grant Execute/SELECT Permissions For All User Defined Functions To A User

You want to add a new user with read and write access and also the ability to execute all user defined functions but you don't want to make the user a db_owner. The code below will do a GRANT EXECUTE/SELECT for all the user defined functions in the DB If the user defined function is a table-valued function then you need to grant select permissions otherwise you need to grant execute permissions

Right now this code prints the GRANT EXECUTE/SELECT statements, change the PRINT to EXEC if you want it to be done automatically

--Grab all the functions for the current DB
SELECT IDENTITY(INT,1,1) AS ID,
SPECIFIC_NAME,DATA_TYPE
INTO #FunctionList
FROM INFORMATION_SCHEMA.ROUTINES --Only Procs
WHERE OBJECTPROPERTY(OBJECT_ID(SPECIFIC_NAME),'IsMSShipped') =0
AND ROUTINE_TYPE='FUNCTION'
ORDER BY SPECIFIC_NAME

DECLARE
@Loopid INT,
@MaxId INT,
@UserName VARCHAR(50)


--This is the user that will get the execute/select permissions
SELECT @UserName = 'SomeUser'


--Grab start and end values for the loop
SELECT @Loopid = 1,
@MaxId = MAX(ID)
FROM #FunctionList

DECLARE
@SQL VARCHAR(500),
@ProcName VARCHAR(400) ,
@Permission VARCHAR(20),
@DataType VARCHAR(20)


--This is where the loop starts
WHILE @Loopid <= @MaxId BEGIN

--grab the function name and type
SELECT @ProcName = SPECIFIC_NAME, @DataType =DATA_TYPE
FROM #FunctionList
WHERE ID = @Loopid

--Find out if it's a table-valued function
IF @DataType ='TABLE'
SELECT @Permission ='SELECT'
ELSE
SELECT @Permission ='EXECUTE'


--construct the statement
SELECT @SQL = 'GRANT ' + @Permission +' ON [' + @ProcName + '] TO ' + @UserName
PRINT (@SQL) --change PRINT to EXECUTE if you want it to run automatically

--increment counter
SET @Loopid = @Loopid + 1
END

--clean up
DROP TABLE #FunctionList

Thursday, April 13, 2006

Grant Execute Permissions For All Stored Procedures To A User

You want to add a new user with read and write access and also the ability to execute all stored procedures but you don't want to make the user a db_owner.
The code below will do a GRANT EXECUTE for all the procedures in the DB
This line will skip those dt_ procedures that are in every database
WHERE OBJECTPROPERTY(OBJECT_ID(SPECIFIC_NAME),'IsMSShipped') =0

Right now this code prints the GRANT EXECUTE statements, change the PRINT to EXEC if you want it to be done automatically

--Grab all the procedures for the current DB
SELECT IDENTITY(INT,1,1) AS ID,
SPECIFIC_NAME
INTO #Procedurelist
FROM INFORMATION_SCHEMA.ROUTINES --Only Procs
WHERE OBJECTPROPERTY(OBJECT_ID(SPECIFIC_NAME),'IsMSShipped') =0
AND ROUTINE_TYPE='PROCEDURE'
ORDER BY SPECIFIC_NAME

DECLARE
@Loopid INT,
@MaxId INT,
@UserName VARCHAR(50)


--This is the user that will get the execute permissions
SELECT @UserName = 'SomeUser'


--Grab start and end values for the loop
SELECT @Loopid = 1,
@MaxId = MAX(ID)
FROM #Procedurelist

DECLARE
@SQL VARCHAR(500),
@ProcName VARCHAR(400)


--This is where the loop starts
WHILE @Loopid <= @MaxId BEGIN

--grab the procedure name
SELECT @ProcName = SPECIFIC_NAME
FROM #Procedurelist
WHERE ID = @Loopid

--construct the statement
SELECT @SQL = 'GRANT EXECUTE ON ' + @ProcName + ' TO ' + @UserName
PRINT (@SQL) --change PRINT to EXECUTE if you want it to run automatically

--increment counter
SET @Loopid = @Loopid + 1
END

--clean up
DROP TABLE #Procedurelist

Wednesday, April 12, 2006

Use OBJECT_DEFINITION To Track Procedure Changes

Not everyone uses Visual Source Safe, CVS or Subversion to keep track of proc changes/deletions and/or additions
Of course you could use Red-Gate SQL Compare (I do) But let's say you don't have any of these tools and are using SQL Server 2005, what else can you do?
In SQL Server 2000 you can use
select ROUTINE_DEFINITION,SPECIFIC_NAME FROM INFORMATION_SCHEMA.ROUTINES to get the body of the stored procedure, the caveat is that this will only return 4000 characters.
Another way is to use the sp_helptext procedure
In SQL server 2005 this is much easier. There is a new function in town: OBJECT_DEFINITION()

OBJECT_DEFINITION() does return the whole body of a stored procedure
Below is some code (very simple) that will give you an idea of how you could use OBJECT_DEFINITION() to keep track of changes
You will have to setup a job that runs once a day and stores the definition of all the procedures in a table
Then you can do a self join on that table to find added, deleted and changed procedures
You can run the code below in 1 shot if you want

USE master
GO
--Let's Create a New Database
CREATE DATABASE TestProcCode
GO

USE TestProcCode
GO


--proc0
CREATE PROC proc0
AS
SELECT
CURRENT_TIMESTAMP
GO

--proc1
CREATE PROC proc1
AS
SELECT
GETDATE()
GO


--proc2
CREATE PROC proc2
AS
SELECT
HOST_NAME()
GO

--create the proc changes table, bad name I know
CREATE TABLE ProcChanges (ID INT IDENTITY,RunDate DATETIME,ProcName VARCHAR(100),ProcCode VARCHAR(MAX))
GO

--insert all the procs that exist now
INSERT INTO ProcChanges
SELECT '20060410',SPECIFIC_NAME,OBJECT_DEFINITION( OBJECT_ID(SPECIFIC_NAME))
FROM INFORMATION_SCHEMA.ROUTINES
GO


--Let's change proc2
ALTER PROC proc2
AS
SELECT
HOST_ID()
GO

--proc 3 is new
CREATE PROC proc3
AS
SELECT
SUSER_SNAME()
GO

--proc 1 is deleted
DROP PROCEDURE proc1
GO

--insert all the procs that exist now
INSERT INTO ProcChanges
SELECT '20060411',SPECIFIC_NAME,OBJECT_DEFINITION( OBJECT_ID(SPECIFIC_NAME))
FROM INFORMATION_SCHEMA.ROUTINES
GO



--grab all deleted procs
SELECT 'Deleted',p1.*
FROM ProcChanges p1
LEFT OUTER JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
AND p2.RunDAte ='20060411'
WHERE p1.RunDAte ='20060410'
AND p2.ID IS NULL


--grab all added procs
SELECT 'Added',p1.*
FROM ProcChanges p1
LEFT OUTER JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
AND p2.RunDAte ='20060410'
WHERE p1.RunDAte ='20060411'
AND p2.ID IS NULL


--grab all changed procs
SELECT 'Changed',p1.*
FROM ProcChanges p1
JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
WHERE p1.RunDAte ='20060410'
AND p2.RunDAte ='20060411'
AND p1.ProcCode <> p2.ProcCode

--grab all procs that didn't change
SELECT 'Not Changed',p1.*
FROM ProcChanges p1
JOIN ProcChanges p2 ON p1.ProcName =p2.ProcName
WHERE p1.RunDAte ='20060410'
AND p2.RunDAte ='20060411'
AND p1.ProcCode = p2.ProcCode


USE MASTER
GO


--let's clean up this mess ;-)
DROP DATABASE TestProcCode
GO


What I have shown is very simple, you can expand on this and check for date ranges and improve on this a lot if you need to

Monday, April 10, 2006

NULL Is Not 'NULL'

Today someone posted a question on the Tek-Tips Forums web site
The queston was how not to insert rows with a NULL value in a certain column
The answer is of course INSERT INTO Table2 SELECT * FROM table WHERE column IS NOT NULL
This person replied that rows where the column is null are still being inserted
I turned out that there was data that had NULL character values 'NULL'
When you run SELECT NULL,'NULL' in Query Analyzer or SQL Server Managment Studio this looks identical so it's very easy to think that there is something else going on

Run the code below in Query Analyzer to understand what I mean

CREATE TABLE #test (
SomeField VARCHAR(50))

INSERT INTO #test
VALUES (NULL)

INSERT INTO #test
VALUES ('NULL')

SELECT *
FROM #test

SELECT *
FROM #test
WHERE SomeField IS NOT NULL

SELECT *
FROM #test
WHERE SomeField = 'NULL'

DROP TABLE #test

Friday, April 07, 2006

SQL Server Everywhere Edition <> Sybase SQL Anywhere

Paul Flessner Senior Vice President of Microsoft Corporation has posted a SQL Server 2005 update here
Am I the only one who thinks that the name "SQL Server Everywhere Edition" is very close to "Sybase SQL Anywhere"?
I can see the jokes already "Not only does SQL server have the same codebase (not true) the name is also ripped off"

Anyway these are the 4 key themes that Paul Flessner mentioned
Continuous Availability and Automation
Beyond Relational
Dynamic Applications
End-To-End Insight

This is what Paul has to say about SQL Server Everywhere Edition "This new offering for storage on clients of all types will provide a lightweight, compact, but rich subset of the capabilities found in other SQL Server editions. Beyond having rich local data management capabilities, SQL Server Everywhere Edition will also include support for seamlessly synchronizing with other SQL Server editions and provides features that promote building rich client applications that operate effectively in today’s increasingly “occasionally connected” environment. SQL Server Everywhere Edition also shares a common programming model with the other SQL Server editions, enabling developers to transfer skills and knowledge quickly and easily. We expect to ship the first CTP of SQL Server Everywhere Edition this summer, with the goal of final release before the end of this calendar year."

Wednesday, April 05, 2006

Left Join

As promised here is a left join blog post. What is a left (outer) join? This is what Books On Line has as the left join description “The Left Outer Join logical operator returns each row that satisfies the join of the first (top) input with the second (bottom) input. It also returns any rows from the first input that had no matching rows in the second input. The nonmatching rows in the second input are returned as null values. If no join predicate exists in the Argument column, each row is a matching row.”
So let’s see how that works

CREATE TABLE #Customer
(CustomerID INT ,
LastName VARCHAR(100),
FirstName VARCHAR(100))

CREATE TABLE #Order
(OrderID INT identity ,
Amount DECIMAL(12,3),
CustomerID INT)

INSERT INTO #Customer VALUES (1,'Gates','Bill')
INSERT INTO #Customer VALUES (2,'Woz','Steve')
INSERT INTO #Customer VALUES (3,'Ellison','Larry')

INSERT INTO #Order VALUES (12.99,1)
INSERT INTO #Order VALUES (13.45,1)
INSERT INTO #Order VALUES (56.45,3)

--Regular Join, CustomerID 2 is not returned
SELECT c.*,o.*
FROM #Customer c
JOIN #Order o ON c.CustomerID =o.CustomerID

--Left Join, CustomerID 2 is returned
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID

--Right Join, CustomerID 2 is returned
SELECT c.*,o.*
FROM #Order o
RIGHT JOIN #Customer c ON c.CustomerID =o.CustomerID

--Return all the people without an order
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
WHERE o.CustomerID IS NULL

--Left Join, CustomerID 2 is not returned because the condition is in the WHERE clause
--The WHERE clause is applied to the #Order table instead of the #Customer table
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
WHERE o.CustomerID < 3

--Left Join, CustomerID 2 is returned (as well as 3) because the condition is in the AND clause
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
AND o.CustomerID < color="#009900">


--The 'correct' way, applying the WHERE condition to the #Customer table
SELECT c.*,o.*
FROM #Customer c
LEFT JOIN #Order o ON c.CustomerID =o.CustomerID
WHERE c.CustomerID < 3


DROP TABLE #Order,#Customer

Tuesday, April 04, 2006

Top SQL Server Google Searches For March 2006

These are the top SQL Searches on this site for the month of March. I have left out searches that have nothing to do with SQL Server or programming. Here are the results...

ToBase64String sql procedure
type of database backup
how to view the picture files for a particular date using sql command
Crystal reports
SQL group right
SQL SERVER create table select
SQL MID
i need to know about money laundry and bluk currency smuggling
AMD SQL Server Performance
sql monitor
convert mysql mssql
trigger
left join SQL Server
left join

It's always interesting to see what people are searching for, I left the money laundry one in this list because I thought it was kind of amusing
I always like to look at this list so that I know what people are interested in and I can write a little thing about it
So tomorrow I will cover LEFT JOIN

Top 5 SQL Server Posts for March 2006

Below are the top 5 posts according to Google Analytics for the month of March

Login failed for user 'sa'. Reason: Not associated with a trusted SQL Server connection. SQL 2005
SQL Server 2005 Free E-Learning
SQL Query Optimizations
Feature Pack for Microsoft SQL Server 2005
Find all Primary and Foreign Keys In A Database

SQL Server 2005 Distilled Book Published

Addison Wesley Professional has published their latest SQL Server 2005 book "SQL Server 2005 Distilled"

Book Description

Need to get your arms around Microsoft SQL Server 2005 fast, without getting buried in the details? Need to make fundamental decisions about deploying, using, or administering Microsoft’s latest enterprise database?



Need to understand what’s new in SQL Server 2005, and how it fits with your existing IT and business infrastructure? SQL Server 2005 Distilled delivers the answers you need–quickly, clearly, and objectively.



Former SQL Server team member Eric L. Brown offers realistic insight into every significant aspect of SQL Server 2005: its new features, architecture, administrative tools, security model, data management capabilities, development environment, and much more. Brown draws on his extensive experience consulting with enterprise users, outlining realistic usage scenarios that leverage SQL Server 2005’s strengths and minimize its limitations. Coverage includes

Architectural overview: how SQL Server 2005’s features work together and what it means to you
Security management, policies, and permissions: gaining tighter control over your data
SQL Server Management Studio: Microsoft’s new, unified tool suite for authoring, management, and operations
Availability enhancements: online restoration, improved replication, shorter maintenance/recovery windows, and more
Scalability improvements, including a practical explanation of SQL Server 2005’s complex table partitioning feature
Data access enhancements, from ADO.NET 2.0 to XML
SQL Server 2005’s built-in .NET CLR: how to use it, when to use it, and when to stay with T-SQL
Business Intelligence Development Studio: leveraging major improvements in reporting and analytics
Visual Studio integration: improving efficiency throughout the coding and debugging process
Simple code examples demonstrating SQL Server 2005’s most significant new features


Table of Contents

Preface xiii
Acknowledgments xv
About the Author xvii


Chapter 1: Introduction to SQL Server 2005 1
Chapter 2: What Everyone Should Know About Security 41
Chapter 3: Enterprise Data Management 83
Chapter 4: Features for Database Development 145
Chapter 5: Overview of Business Intelligence 197
Chapter 6: The Code Chapter 245

Appendix A: SQL Server 2005 System Information 285
Appendix B: System Tables and View in SQL Server 2005 291
Appendix C: SQL Server Built-In Functions 295

Index 297


The Amazon link is here

Tuesday, March 28, 2006

Use sp_executesql Or EXEC To Get The Count For Dynamic Table Names

You need the record count for a table but this table could change every day or change based on which user executes the stored procedure
There is no point in creating dozens of stored procedures, all this can be done by using exec (sql) or sp_executesql
Of course sp_executesql is much better and if you run the statements below you will also see that it's execution plan is only 29.04 percent relative to the batch
So let's get started with sp_executesql

USE pubs
GO

--sp_executesql
DECLARE @chvTableName VARCHAR(100),
@intTableCount INT,
@chvSQL NVARCHAR(100)

SELECT @chvTableName = 'Authors'
SELECT @chvSQL = N'SELECT @intTableCount = COUNT(*) FROM ' + @chvTableName

EXEC sp_executesql @chvSQL, N'@intTableCount INT OUTPUT', @intTableCount OUTPUT

SELECT @intTableCount
GO


--EXEC (SQL)
DECLARE @chvTableName VARCHAR(100),
@intTableCount INT,
@chvSQL NVARCHAR(100)


CREATE TABLE #temp (Totalcount INT)
SELECT @chvTableName = 'Authors'
SELECT @chvSQL = 'Insert into #temp Select Count(*) from ' + @chvTableName

EXEC( @chvSQL)

SELECT @intTableCount = Totalcount from #temp

SELECT @intTableCount

DROP TABLE #temp


First Hit CTRL + K (this will show the execution plan) , then highlight the complete code, hit F5 and look at the Execution Plan tab
As you can see sp_executesql is more than twice as efficient as exec (SQL)