user-specified point).
The program
-\W{http://www.chiark.greenend.org.uk/~sgtatham/agedu}\cw{agedu},
+\W{http://www.chiark.greenend.org.uk/~sgtatham/agedu/}\cw{agedu},
which uses a data structure of this type for its on-disk index
files, is used as a case study.
structure, but the resulting structure is not dynamically updatable
in turn, it follows that this technique cannot be applied twice: no
analogous transformation will construct a \e{three}-dimensional
-struccture capable of counting the total weight of an octant
-\cw{\{x\_ <\_x0, y\_<\_y0, z\_<\_z0\}}. I know of no efficient way
+structure capable of counting the total weight of an octant
+\cw{\{x\_<\_x0, y\_<\_y0, z\_<\_z0\}}. I know of no efficient way
to do that.
The structure as described above uses \cw{O(N log N)} storage. Many
\C{application} An application: \cw{agedu}
The application for which I developed this data structure is
-\W{http://www.chiark.greenend.org.uk/~sgtatham/agedu}\cw{agedu}, a
+\W{http://www.chiark.greenend.org.uk/~sgtatham/agedu/}\cw{agedu}, a
program which scans a file system and indexes pathnames against
last-access times (\q{atimes}, in Unix terminology) so as to be able
to point out directories which take up a lot of disk space but whose