Although with my collation change, I did it within the database, probably the next time I would test if copying all the data to a database with a 'correct' collation would be fast enough and use that.

I did have a look in the code I used before, see some notes below.

If anybody has a better (alternate) method, please inform us !

Succes,

Ben

Alltogether I performed the 'conversion' in 8 steps.

step 0.

Generate the script to create complete database. Remove all creation of the tables.

(This script is used in step 7).

Step 1.

Alter the database collation.

Step 2.

Check a number of occurences. (No changes in thet database).

Step 3.

Create a table which registers the DOMAIN_USAGE.

(Where are the user defined types used).

Step 4.

Removing triggers, constriants, indexes etc.

a. ('F','TR') sysobjects

b. ('C', 'PK', 'UQ', 'D' ) sysobjects

c. sysindexes.name like 'IX_%'

Step 5.

Chance all the datatypes to the correct collation.

Step 6.

Restore all DOMAIN_USAGE fields (use the table created in step 3).

Step 7.

Run the script prepared in Step 0.

The Process workt, but it was not all happy flow. (This was a number of years back, and a much older version of SQL-server)

So now I would be inclined to Create a 'new' database with the correct collation and move all data to that database. (Probably removing the indexes before and reinstalling them after the move).