James Bonfield - 2011-04-18

Looking at this code it appears a bit dumb at present. If your read doesn't contain a dot then it assumes it lacks suffixes and fwd/rev pairs will have the same name. Then it just appends ".<record_id>".

Ideally it'd spot duplicates and resolve the conflicts, but it may be very slow to implement. I'll experiment. I've already found a related bug as contig names can clash with reading names too.