i was wondering about something that is in dc++... if you download something and want to "match quete" form a user if the files are the same but named diffrently it wont work
did you consider that dc++ would create a file with md5 sums of shared files and send it with the file fist...? if the matching would be done only by size and sums it would find and add more sources (i think)...?
also searching for a specific md5 sum on the hub would make the acuracy of findings less sensitible to file names + less wrong hits