well your suggestions are great, i had the recursive function calls as my first solution during the research but i used this solution from Hunterwood

;with HierarchyTree ([HierarchyKey], [ParentKey], [Level], nodePath)

as

(

...

good solution, but assuming the Depth/Level is infinite, creating an uncertain..

[HierarchyKey] as [Hierarchy!2!HierarchyKey],

ParentKey as [Hierarchy!2!ParentKey],

[Level] as [Hierarchy!2!Level],...

bump, anyone have more idea?

GSquared (9/4/2008)

---

jj, that still doesn't get it. You have to know the range before-hand for that solution to work. If you know the range already, then you don't...

GSquared (9/4/2008)

---

jj, the problem with that is it has to be a duplicate, since the last duplicate.

Getting all the duplicates is easy. Getting the duplicates since the last duplicate...

hmm... interesting, how about having a counter that resets to 1 when it encounters a duplicate... i'm not sure if this is applicable/achievable using partitioning/ranking,

my thought would be

RowIDRowKeyCounter

151

251

342

433

524

611

711

Then selecting all...

hmm i have not actually dug deeper with bcp, can we monitor the records being inserted into the table? like error in each fields in a row, fk_constraints, pk_constraints, Null...

yes, but in case it has multiple result, it will yield to the first non null column, so even col1 could be updated from row 1 in results matching to...

never mind found the answer,

SQL behaviour would be...

First Non Null Field for Each group of Record would be updated to the source...

so with the tables above this would mean, only

Table...