Usage:
    cat xy.csv | sum-distinct 

I use this script when counting the response times of a web service
per minute. Often I take the access logs, sed/awk/cut the time of
day and the milliseconds so that my "x" is the HH:MM and the "Y" is
the duration. When done, I have the total duration per minute 
(sorted lexically).

It's perl as I often need to bring this to new servers which often
are "bare bones".