Lynn Pettis (6/6/2010)

---

Steve,

You really are missing the entire point. I am not disputing what the function does but HOW it does it. Your use of a recursive CTE...

Lynn Pettis (6/6/2010)

---

Steve,

Unfortunately, you are using the same logic everyone else uses to defend the use of inefficient code, "It works for my case." Problem is, it is still...

Martin Mason (6/4/2010)

---

Hey. Why stop there! If separate date/time dimensions leads to the same number of possible cells in the resultant cube space as the a single date/time dimension, that...

But to combine date and time just increases the sparcity of the multidimensional database along a dimension that is most likely going to be used to partition a database of...

That's a matter of perspective, and not a matter of fact. Under what circumstance is a day not subdivision of some constantly distributed...

Not having "29 years of experience" but being keenly interested in learning from those with good ideas &/or years of experience/expertise, I am disappointed when I see comments such as...

Martin Mason (6/2/2010)

Whoa. Timeout. If you're talking about SSAS, your argument only considers partition processing, NOT dimension or aggregation processing SPACE, TIME, or MEMORY CONSUMPTION. With separate dimensions, the time...

That is quite a strange response. Your article spent time basically explaining how your Gregorian calendar approach using a recursive CTE is the only "good" solution. (Afterall, you...

I'd already read it when it was first posted to the thread - but when...

Lynn Pettis (6/2/2010)

---

What most people are saying is that this approach isn't necessarily the correct approach in a DW environment.

My complaint with the code is the use of a recursive...

Not sure you're addressing the central issue. It is almost always better to separate the date and time dimension. I add almost not because of can think of an exception,...

Whilst there may be more efficent mechanisms to generate the data itself (you may get better with a loop in a stored procedure or other evils), it's a lot easier...

Just noticed I'd posted to the SQL 2K forum from your comment. Thats what I get for mis-reading the Active Topics list 😉

I mean, rather than trying to figure out what's wrong with the SQL, if you wrote some queries to join the tables together and see why they're not being deleted,...

Fairly straightfoward - use mated ROW_NUMBER() rankings, one flowing forward and one flowing backward over the data. By taking the first record where the orders cross over, you can get...