There is also the school of thought that we ought to identify
what problem we are trying to solve, and if it's merely a
'search/browse' type of problem, then we should fix our
search/browse interfaces and leave the directory layout alone.
Unfortunately, this school of thought seems to have only one
adherent, which is myself :-)
Coung mr in.
It occurs we may be looking at multiple "problems", which may
or may mot have compatible solutions. Someone mentioned finding a
(popular) port with a known name; I often want to look at all
programs which do <foo>, where FOO is far more detailed than the
existing categories support.