It depends whether or not you want to convert all the times so that they are relative to the same timezone and are therefore more easily comparable. If so, and...

Jacob Pressures (5/14/2009)

---

WEll I've figured out why it is turning it NULL because I'm not returning anything.

I should have and ELSE that simply returns EmergencyRelation. But i still don't know...

This ROW_NUMBER() difference method used to solve the problem of combining sequential rows originated from the following article, or at least that's where I first saw it used.

http://www.sqlmag.com/Article/ArticleID/93606/sql_server_93606.html

Suppose we have...

The datetime data type is stored internally as two 4-byte integers concatenated together rather than a true floating point type.

The first 4-byte integer represents the date portion of the...

I couldn't reproduce your "joined" data when I ran the following query against your previously supplied base table data (see post on 17th April).

SELECT S.Start, S.Stop, L.PLCount

FROM dbo.LTrack L

...

Here's an alternative method that uses CTE's. It produces the same results as Wayne's quirky update given the OP's test data.

ConsumerFK ActiveDays

33 ...

From Books Online:

"CHECKSUM satisfies the properties of a hash function: CHECKSUM applied over any two lists of expressions returns the same value if the corresponding elements of the two lists...

Here's an alternative query that uses a method similar to Mike McQueen's to parse the times from the time strings when converted to integers. This method does have the advantage...

Zeev Kazhdan (4/27/2009)

---

Thanks, but adding of

UNION ALL

SELECT 1,6,100,1,100,15

my example will return zero duration for this Top record. This is a place to fix my explanation's mistake - in case...

I think your problem is solvable using a HAVING clause which ensures that the number of records for each user_id that is returned is equal to the number of model_id...

I'm not sure if the following really does what you need, but it does reproduce your expected results.

;WITH scheduled_commercials AS (

SELECT 1 AS commercial_break_id,

...

You can't modify data in base tables from a user defined function.

If you want to reuse this TSQL you could consider putting it in a stored procedure.

By converting the time strings to datetime values, you can do the following.

/* I'm assuming here that @TimeString1 always represents the earlier time and @TimeString2 the later time */

DECLARE @TimeString1...

You could just replace every occurrence of compdte in your query with

ISNULL(NULLIF(compdte, 0), GETDATE())

A datetime value of '1900-01-01 00:00:00.000' is 0 (zero) when the datetime is converted to...

Sorry but I still don't understand your requirements, such as which columns should be counted and which columns should be grouped / binned. Rather than making a guess as to...