Distinct on more then 1 column

I know this sounds a bit silly, but as of now I need a solution. Say if I have a table:

userid |开发者_如何学编程 category  
-------+---------  
derkv2 | Batch  
markj  | HTFS  
marjk  | TERMK  

How can I return unique rows, only take the first category the user belongs to? So it will look like:

userid | category  
-------+---------  
markj  | HTFS  
derkv2 | BATCH

You must fully identify each row that you want to retrieve. Assuming that your tables are well-crafted and have a unique PK then you could do something along these lines:

select
  *
from
  table
where
  pk in (select min(pk) from table group by userid)

I would use a partition. Let's you aggregate without ripping apart table. Add the partition to your source table and it becomes a unique contraint in the process.

    select col1, 
           col2, 
           row_number() OVER (Partition BY col1 order by col2) ID
           From dbo.Example

Results

--x--|--y--

--A--|--1--

--A--|--2--

--A--|--3--

--B-|--1--

--C--|--1--

--C--|--2--

etc