Update statement using function

Question

Environment: Win7 and SQL server 2008 R2
Tools: SQL management tool 2008 R2
Problem: I have been trying to update id numbers in the staging table. T-SQL statement updates all id number in the staging table, but what if the we have multiple records for the same person who we are assigning the ID to his/her record. It should generate
one pn_id per ssn regadless if there are several records for the same individual. for example, my code assigns the following id: 
Student table (ID, ssn, name, subject, grade, year)
(288258466,192160792 , Sally Johnson, Math, A, 2014 )
(176268917, 192160792, Sally Johnson, Spanish, B+, 2014 )
(900303787, 770616858, Jessica Simpson, Spanish, B, 2014 )
(704099635, 770616858, Jessica Simpson, IT, A, 2014 )
(704099640, 444718562, Douglas Mike, IT, A, 2014 )
(288258450, 31883459, Susan Jesson, IT, A, 2014 )
(176268960, 260518681, Veronica Floris, IT, A, 2014 )
The expected results should be as the following in the staging table
Student table (ID, ssn, name, subject, grade, year)
(288258466,192160792 , Sally Johnson, Math, A, 2014 )
(288258466, 192160792, Sally Johnson, Spanish, B+, 2014 )
(900303787, 770616858, Jessica Simpson, Spanish, B, 2014 )
(900303787, 770616858, Jessica Simpson, IT, A, 2014 )
(704099640, 444718562, Douglas Mike, IT, A, 2014 )
(288258450, 31883459, Susan Jesson, IT, A, 2014 )
(176268960, 260518681, Veronica Floris, IT, A, 2014 )
My code:
UPDATE a
SET pn_id = (Select dbo.gen_id())
from [dbo].[Staging_Table] a
where SSN in (Select SSN from [dbo].[staging_Table]
group by SSN having count(SSN) > 1)
GO
I also tried the following code but no success
;with cte
As
Select * , ROW_NUMBER() Over(Partition BY ssn Order by ssn) As MyCount
from dbo.Staging_Table
Select * into #a from cte where MyCount=1
UPDATE a
SET pn_id = (Select dbo.fn_gen_id())
from staging_table a
Full join #a b on a.ssn = b.ssn
Drop table #a
please help.

Answer 1

I ran some tests and Sandra is correct, you don't always get the same value for all occurrences of the same SSN.
After thinking about this, I think I understand why this happens.
With functions, SQL is allowed to assume that multiple calls to a function in a single command with the same parameter values always return the same result.  That's why functions cannot use side-effecting functions and is not allowed to make changes
to any table except table variables declared in the function.
And SQL is allowed to rearrange the processing of a command in any manner it believes is most efficient as long as it returns the same result.  So even though the query I gave says first get a set of distinct SSN's and then calls the function only once
for each SSN, it is valid for SQL to instead call the function once for each row in the original table since this function has no parameters and is therefore should return the same value on each call.  But since the function being called apparently (we've
not seen the code of the function) returns different values on different calls, the same SSN could be assigned different values, but SQL would still be working correctly.
Note that it would also be valid for SQL to notice that this function has no parameters and therefore should return the same value every time it is called within the same command and so just call the function once per execution of the command.  This
would mean that every SSN would get the same ID.  As far as I can determine currently SQL does not do that but a different version in the future with a different optimizer might.
Tom