Modularization for the KWIC System
==================================
as resulting from the discussion in the lectures on Nov. 26 / Dec. 3, 2002

first modularization
====================

input
-----
interfaces:
- in:  lines of characters
- out: array/list of lines of characters, end-of-word markers
  - take care that data format suits the needs of all the following modules
sub-modules:
- distinguish word boundaries
- check for legal input

permutations
------------
interfaces:
- in:  array/list of lines of characters, end-of-word markers
- out: a permuted array/list of ...

sort
----
interfaces:
- in: (same as above)
- out: (same as above, but sorted)

output
------
interfaces:
- in: (same as above)
- out: printout/file with printout, formatted nicely
sub-module:
- inverse of: distinguish word boundaries


question:
-.-.-.-.-
- are these write-time modules?


alternative modularization
==========================
input and output
----------------
sub-modules:
- ...
interfaces:
- in:
- out:

permutation and sorting
-----------------------
sub-modules:
- ...
interfaces:
- in:
- out:

reason for this choice:
-.-.-.-.-.-.-.-.-.-.-.-
- input and output are related (people are usually experts for both)
- permutation and sorting are related (people are usually experts for both)

problem:
-.-.-.-.
- the interface of the "input and output" module is rather big and complex


criteria for a good modularization
==================================
- create a write-time modularization
  - assign to different people
- separate different functionality
- we want to use separate people/experts for writing "input" and "output"
- have tiny but precisely specified interfaces
- interfaces between modules should match