Achim Gratz <Stromeko at nexgo.de>: > Eric S. Raymond writes: > > OK, there are two ways we can handle this. > > The third way is building a hash table that maps each possible unique > prefix to the actual full-length command. That seems like a detail in how the second approach would be implemented. -- <a href="http://www.catb.org/~esr/">Eric S. Raymond</a>