-- ... query runs here and returns row

IF @@ROWCOUNT = 0 RAISERROR('No records found', 18, 1)
ELSE IF @@ROWCOUNT > 0 RAISERROR('More than one record found', 18, 1)

-- ... continue when exactly one row is returned

Now if your query returns 3 records you’d expect it to trip the second IF statement and raise error with correct message. Unfortunately this doesn’t happen. The reason being – the first IF statement resets @@ROWCOUNT to 0. So, to avoid this we need to preserve the @@ROWCOUNT value in a local variable:

DECLARE @iRowCount int

-- ... query runs here and returns row

SET @iRowCount = @@ROWCOUNT

IF @iRowCount = 0 RAISERROR('No records found', 18, 1)
ELSE IF @iRowCount > 0 RAISERROR('More than one record found', 18, 1)

-- ... continue when exactly one row is returned

This way count of rows returned by the query is saved and is not affected by any following statements.

2011-08-19	16
2011-08-22	45
2011-08-24	62
2011-08-25	88
2011-08-27	17
2011-08-28	35
2011-09-01	10
2011-09-02	79
2011-09-03	70
2011-09-07	83

As you can see not all dates are in the sequential order, 24th comes after 22nd etc. But in many cases you need this data to be sequential, for example if this data feeds a chart you need to fill it with missing dates and 0 for a value. There is an easy way to achieve this.

Some solution suggest creation of a Calendar table – table with prefilled dates and values, but for this task it would be an overkill. A small Common Table Expression will suffice instead.

Let’s say your table with data is called #TEMP_DATA, column with dates is called MyDate, column with integer values is called MyValue. Consider following CTE:

-- Getting date limits
DECLARE @MaxDate DateTime
DECLARE @MinDate DateTime

SELECT @MaxDate = MAX(MyDate), @MinDate = MIN(MyDate) FROM #TEMP_DATA;

-- Calendar CTE with ALL the dates
WITH mycte AS
    (
        SELECT @MinDate AS DateValue
        UNION ALL
        SELECT DateValue + 1
        FROM mycte
        WHERE DateValue + 1 <= @MaxDate
    )
SELECT DateValue, 0 AS IntValue
FROM mycte
OPTION (MAXRECURSION 0)

It will produce full date range within limits specified by dates in #TEMP_DATA table:

2011-08-19 	0
2011-08-20 	0
2011-08-21 	0
2011-08-22 	0
2011-08-23 	0
2011-08-24 	0
2011-08-25 	0
2011-08-26 	0
2011-08-27 	0
2011-08-28 	0
2011-08-29 	0
2011-08-30 	0
2011-08-31 	0
2011-09-01 	0
2011-09-02 	0
2011-09-03 	0
2011-09-04 	0
2011-09-05 	0
2011-09-06 	0
2011-09-07 	0

Only thing remains is to marry it in happy UNION to original table with data:

DECLARE @MaxDate DateTime
DECLARE @MinDate DateTime

SELECT @MaxDate = MAX(MyDate), @MinDate = MIN(MyDate) FROM #TEMP_DATA;

-- Calendar CTE with ALL the dates
WITH mycte as
    (
        SELECT @MinDate AS DateValue
        UNION ALL
        SELECT DateValue + 1
        FROM mycte
        WHERE DateValue + 1 <= @MaxDate
    )
-- Original data
SELECT MyDate, MyValue FROM #TEMP_DATA
UNION
-- Filled in data
SELECT DateValue, 0
FROM mycte LEFT JOIN #TEMP_DATA ON DateValue = MyDate WHERE MyDate IS NULL
ORDER BY 1
OPTION (MAXRECURSION 0)

Query selects original data that has dates missing and then UNIONs it with full date range from the CTE which is limited only to missing dates by LEFT JOIN. The result is perfect combination:

2011-08-19	16
2011-08-20	0
2011-08-21	0
2011-08-22	45
2011-08-23	0
2011-08-24	62
2011-08-25	88
2011-08-26	0
2011-08-27	17
2011-08-28	35
2011-08-29	0
2011-08-30	0
2011-08-31	0
2011-09-01	10
2011-09-02	79
2011-09-03	70
2011-09-04	0
2011-09-05	0
2011-09-06	0
2011-09-07	83

All the dates are in sequential order, original date range is preserved and missing values are filled with zeroes.

ALTER PROCEDURE Proc1(@val1 NVARCHAR(max))
AS
BEGIN
    -- some code
    EXEC Proc2 @val2 = @val1
    -- some other code
END

I was getting error “String or binary data would be truncated”, but only when 2 conditions were met:

1. @val1 is quite large
2. Either SQL Server/computer just restarted or stored procedure Proc2 was just updated (ALTER)

Error happened just by the fact of the EXEC Proc2 being there it didn’t even had to do anything, it could RETURN straight away. Both @val2 an @val1 are of a type NVARCHAR(max) so there is no reason for the error. The error happened only once, after that I can pass data of any size – and no error would happen. Like I said – nuts.

Maybe it’s a bug in SQL Server – it doesn’t expect such a large value from start, maybe I am doing something wrong. The solution I came up with is kinda weird, but it works: On the very fist call to Proc1 I pass a very small predefined value for @val1 parameter, something like !TEST!. The call goes without any error and and all consecutive calls with any data size as well. The only thing Proc2 has to detect that !TEST! value is passed and don’t do any actual work with it, just RETURN straight back

You’re calling SQL Server stored procedure from your .NET code and it runs extremely slow. When you run same SP with exactly the same parameters (as captured by SQL Server Profiler) directly in SQL Server Management Studio, it runs very fast. What gives?

Chances are – that SP was executed before and query plan was cached for the specific parameters. To avoid this add WITH RECOMPILE option to your CREATE PROCEDURE or ALTER PROCEDURE statement. This will force SQL Server to create a new query plan every time SP runs, perhaps adding slight overhead, but creating an optimized path that will cover that overhead tenfold.

OraOLEDB.Oracle Provider is not registered on the local machine

Now I know that the driver was installed and registered. I downloaded an official driver from Oracle WebSite. In my case it was ODAC112021Xcopy_x64.zip 64-bit version for XCopy deployment. It installs in 2 easy steps:

1. Unzip downloaded file into any folder
2. Run (as administrator) command: INSTALL TYPE PATH NAME DEPENDANCIES

Where

• TYPE – type of installation (e.g. OLEDB, basic etc.)
• Path – where you want driver installed
• Name – Oracle home name
• Dependencies – true/false whether to install dependencies (e.g. instant client)

So my command was

INSTALL ALL “C:\Program Files\Oracle64Driver” Oracle64Driver TRUE

Which copied the files and created correct Registry entries (I checked). And still I was getting the error. I Googled it (a lot) but majority of suggestions was that the error is due to Windows ACL and correct permissions should be set on the driver folder. Didn’t help me.

So I fired up trusted ProcessMonitor and it showed that W3WP.EXE (ASP.NET process) was trying to access missing OCI.DLL file in the path C:\Program Files\Oracle64Driver\Bin, e.g. in the Bin folder of the path were the driver was installed. Looking back at the place were I unzipped the original driver files I found that DLL inside of “instantclient” folder. So I copied entire content of that folder into Bin folder at the destination. And Voila! The error disappeared.

Apparently Instant Client files aren’t copied by the installer even when Dependencies option is set to true.

ALTER PROCEDURE MyProcedure(MyParam int) ...

after dragging it into Linq2Sql designer you can call it in your .NET code like this:

Dim aResults = MyDbContext.MyProcedure(2011)

but there are 2 caveats.

1. If you plan to use stored procedure in a Linq2Sql query – do not use temp tables. It confuses ORM, it cannot determine type of output result and you will get a warning:

   The return types for the following stored procedures could not be detected

   If you need to store temporary data somewhere – use table variables instead.

2. If you use code above as is and then attempt to use query variable multiple times (e.g. get the total Count of records first and then iterate thru variable in For Each loop or bind it to a control – you will get an error:

   The query results cannot be enumerated more than once

   This happens because underlying stored procedure can be executed only once and result returned by the Linq2Sql method is of type ISingleResult(Of T). The solution is to convert the result to in-memory list:

   Dim aResults = MyDbContext.MyProcedure(2011).ToList()

   This way the result becomes of type Generic.List(Of T) and you’re free to do whatever you need with it as many times as you need.

In order to successfully implement this use case 2 items need to be addressed:

First in the report itself parameters need to be ordered in such way so “filter” comes before “dropdown”. Parameters can easily be arranged in Business Intelligent Development Studio, by expanding Parameters node, selecting a parameter, and using arrow buttons in Report Data menu.

Second in the ASP.NET application, configuring ReportViewer control (setting credentials, server URL, report path and our “filter” parameter) needs to be done in Page_Init event in the “If Not IsPostback” block.

Let’s say you have 2 following queries from Northwind database:

SELECT TOP (5) CompanyName FROM Suppliers ORDER BY CompanyName

Aux joyeux ecclésiastiques
Bigfoot Breweries
Cooperativa de Quesos 'Las Cabras'
Escargots Nouveaux
Exotic Liquids

and

SELECT TOP (5) ProductName FROM Products ORDER BY ProductName

Alice Mutton
Aniseed Syrup
Boston Crab Meat
Camembert Pierrot
Carnarvon Tigers

And now you want to combine them into a single result, adding title lines to separate each result. The direct approach would be:

SELECT 'Suppliers' AS Name
UNION
SELECT TOP (5) CompanyName FROM Suppliers
UNION
SELECT 'Products'
UNION
SELECT TOP (5) ProductName FROM Products
ORDER BY Name

But result is far from what we wanted

Alice Mutton
Aniseed Syrup
Aux joyeux ecclésiastiques
Bigfoot Breweries
Boston Crab Meat
Camembert Pierrot
Carnarvon Tigers
Cooperativa de Quesos 'Las Cabras'
Escargots Nouveaux
Exotic Liquids
Products
Suppliers

Entire resultset is sorted uniformally and products are mixed with suppliers.

The solution is to designate a special “Sort By” field in each query. Consider following revision of previous UNION

SELECT 'Suppliers' AS Name, 1 as SortBy
UNION
SELECT TOP (5) CompanyName, 2 FROM Suppliers
UNION
SELECT 'Products', 3
UNION
SELECT TOP (5) ProductName, 4 FROM Products
ORDER BY SortBy, Name

Here were added a numeric field to each query, 1 to the 1st, 2 to the 2nd, 3 to the 3rd and 4 to the 4th. And then in the ORDER BY clause we sort by this field first and the name second. The result is properly sorted list of suppliers & products:

Suppliers                                   1
Aux joyeux ecclésiastiques                  2
Bigfoot Breweries                           2
Cooperativa de Quesos 'Las Cabras'          2
Escargots Nouveaux                          2
Exotic Liquids                              2
Products                                    3
Alice Mutton                                4
Aniseed Syrup                               4
Boston Crab Meat                            4
Camembert Pierrot                           4
Carnarvon Tigers                            4

WITH DUP_TABLE AS
   (SELECT ROW_NUMBER()
    OVER (PARTITION BY FIELD1, FIELD2 ORDER BY FIELD1, FIELD2) As ROW_NO
    FROM ORIGINAL_TABLE)
DELETE FROM DUP_TABLE WHERE ROW_NO > 1

Here ORIGINAL_TABLE is your table with duplicates. FIELD1 and FIELD2 are columns with duplicates value (feel free to add or remove columns to suit your needs). Internal query assigns a row number to each duplicate record and DELETE statements that uses that CTE deletes all the rows except the one with Row Number = 1

The reason in this case – when you change the data source thru GUI – it resets query used by the report. It can change query type from StoredProcedure to Text and also remove all default query parameters (hence error above).

There’re 2 possible solutions here: First – instead of using GUI, manually change Data Source in the XML source of the report file (in Solution Explorer right mouse click on the report file and select “View Code”). You will have to change it in several places, so be careful. Second – re-create query with its default parameters after changing Data Source in GUI.