As a side-note:

If you use RANK rather than ROW_NUMBER, you will at least always get the same result for the same input.

This is the 'WITH TIES' problem. RANK will...

Not as far as I know, no.

Just a small correction is needed:

vk-kirov (2/18/2010)

---

I am puzzled with the...

Did you see the updated code?

UPDATE ECA

SET memkey = ML.memkey,

state = ML.state,

...

Barry,

Basically, it comes down to the fact that UDTTs are a TYPE.

When you create an instance of the type, you create a new table - which has its own object_id...

Grant Fritchey (2/18/2010)

---

I must stink at my job because I've got BOL open most of the time. I'm just not smart enough to remember everything. Must be why I'm not...

Let me be up-front about this: it's very difficult to say what the cause is from the data presented so far.

But, let me have a crack at it anyway.

Yes, it...

A...

Gianluca Sartori (2/18/2010)

---

Ok, it's not guaranteed and I would not rely on it, but honestly I have to say I've never seen it behaving differently. I've been spending...

GilaMonster (2/18/2010)

---

Other option is to fragment the clustered index and then switch to read-uncommitted or force a table lock (can't recall if it has to be exclusive or if shared...

tony rogerson (2/18/2010)

---

Yes - I do mean a cursor; a static one and not a "update where current".

I don't honestly know what your problem is here Paul - even Jeff...

Gianluca Sartori (2/18/2010)

---

GilaMonster (2/18/2010)

---

Logical order of the index - yes

Physical order of the data...

GilaMonster (2/18/2010)

---

If I'm reading the q correctly, he wants the data in the identity order,...

:blush: ...

I think you're making an unsound logical leap there Tony 😉

tony...