>>Trying to avoid cursors and procedural code in tsql
>>
>>I have a table derived from an SSIS feed that has 1 - n rows that are related, based on symbol, eff_date. If there are more than 1, I want to concatenate the values in a varchar column into the first row
>>
>>1000 rows, may have 200 rows belonging to 60 sets of 2-4 members
>>
>>First row of set and any row that has no set has a Seq_num = 1 ( also have a column fgkey int which acts as a key for a set so can use that instead of symbol, eff_date to find rows belong to a set. Singleton rows have no fgkey )
>>
>>I also have a bit column which I want to be 1 in the first row if any of the rows in the set have a 1
>>
>>So
>>symbol eff_date terms price_sens seq_num
>>MSFT 20120601 'string1' 0 1
>>MSFT 20120601 'string2' 1 2
>>MSFT 20120601 'string3' 0 3
>>
>>would become
>>
>>MSFT 20120601 'string1 , string2 , string3' 1
>>MSFT 20120601 'string2' 1 -- ( this can stay as it is)
>>MSFT 20120601 'string3' 0 -- ( this can stay as it is)
>>
>>Basically it would be an aggregate function but on strings instead of adding numbers.
>>
>>If a sum of a bit column would work for finding if the total > 0 I can work that part out.
>>
>>Looking for some set-based command or magic that would concat the strings
>>
>>TIA
>
>Hi Charles,
>
>I recently had to do something similar (but simpler) for a quick one-off report. Take a look at some of the ideas proposed here
>
>http://stackoverflow.com/questions/194852/concatenate-many-rows-into-a-single-text-string.
>
>The second answer is the one I implemented. I have to admit, I have no idea how it works, but it is very cool.

Thanks Bruce. Off to read up and I'll report back.