It's important to separate the statement of the algorithm from its
implementation.
The algorithm is now stated in two separated phases:
1) Property lookup
2) Action based on properties
This makes the statement of the algorithm much cleaner and easier to verify,
since the part 2) is now independent of adding characters to the standard.
As Mark points out, you can use any number of tricks to get your desired
time-space tradeoff in your implementation.
A./
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:02 EDT