A Coding Challenge From The Blindman.

Hey it's not often the blindman asks for advice on sql coding (never, I think), so here is an opportunity to solve a problem I've been knocking my head against for two days.

Here is sample code for setting up the problem:create table #blindman
(pkey smallint primary key,
fkey char(1),
updateddatetime)

insert into #blindman (pkey, fkey, updated)
select1, 'A', '1/1/2006'
UNION
select2, 'B', '1/1/2006'
UNION
select3, 'A', '1/2/2006'
UNION
select4, 'B', '1/2/2006'
UNION
select5, 'A', '1/4/2006'
UNION
select6, 'B', '1/2/2006'
UNION
select7, 'A', '1/3/2006'
UNION
select8, 'B', '1/3/2006'
UNION
select9, 'A', '1/5/2006'
UNION
select10, 'B', '1/5/2006'

drop table #blindman
Notice that for fkey 'B', there are two entries with '1/2/2006', and for fkey 'A' the updated values are not in synch with the order of the primary key.
The challenge: determine the next pkey for each pkey value, ordered by [updated], and using pkey as a tie-breaker when two records have the same [updated] value.
Here is the desired output for the sample data:pkey fkey updated nextpkey
------ ---- ---------- --------
1 A 2006-01-01 3
3 A 2006-01-02 7
7 A 2006-01-03 5
5 A 2006-01-04 9
2 B 2006-01-01 4
4 B 2006-01-02 6
6 B 2006-01-02 8
8 B 2006-01-03 10
Records 9 and 10 are missing because they have not succeeding records, though I'd be just has satisfied to include them with NULL as their nextpkey value.
Ideally, I want this as a VIEW.

Who's up for the challenge?