Top-Down Design

• Top-Down Design: work from generalities to specifics.
• Bottom-Up Design: work from specifics to generalities.
• Both methods have strong supporters.

Top-Down Design

• Also known as decomposition.
• Begins with a high level of abstraction and build down.
• Iterative method; works using a divide and conquer approach.
  • Only need to focus on details for one part of the system at a time.
  • Decompose top level, then decompose those items, etc.
  • Stop dividing when the decomposition becomes trivial.
• Contrasts with bottom-up design (also known as composition).

Arguments For

• We can only concentrate on so many things as once.
• Defers deciding on low-level specifics which are subject to change.
• More ‘natural’.

Arguments Against

• Problems may be so abstract its hard to get started.
• May not identify critical functionality early on.
  • Unforeseen complexity can ripple back up, making things overly complex.
• Low-level constrains can often dictate higher level design.

Types of Decomposition

Functional Decomposition

• Breaking apart program behaviour (functions) into subprograms, simplifying higher level algorithmic description

Modular Decomposition

• Breaking apart program structure into modules.
• Each has well defined scope of operations (simpler).
• Programs can be expressed in terms of use of and interaction between modules.

Data Decomposition

• Breaking up data structures into components, each of which is simpler and relatively independent of the aggregate data structure.