[ned1] / docs / design_goals.md

Goals
=====

* Build real hardware to maintain interest in software development.
* Sanity check before attempting a workstation-capable machine.
* Develop better hardware intuition.
* Include sufficient capability to handle a real application after completion
  of experimentation. A calculator or instrument controller would be
  satisfactory.
* Establish a performance mark against which to predict/compare future designs.

Principles
==========

* Simplicity.
* Let the implementation guide the definition.
* Keep options open. Postpone decisions where possible.
* Design for debugging and rework. The CPU should be easily instrumented.
* Defining what IS NOT is as important as defining what IS.
* Don't add functionality unless a real application requires it.
* Don't get bogged down in I/O. Use modern hardware where helpful.
* Worse is better. If 90% functionality can be implemented for 10% effort, do
  it and move on.
* The only thing worse than generalizing from a single example is generalizing
  from no example at all.
Commit	Line	Data
bc5b63cf AT	1	Goals
	2	=====
	3
	4	* Build real hardware to maintain interest in software development.
	5	* Sanity check before attempting a workstation-capable machine.
	6	* Develop better hardware intuition.
	7	* Include sufficient capability to handle a real application after completion
	8	of experimentation. A calculator or instrument controller would be
	9	satisfactory.
	10	* Establish a performance mark against which to predict/compare future designs.
	11
	12	Principles
	13	==========
	14
	15	* Simplicity.
	16	* Let the implementation guide the definition.
	17	* Keep options open. Postpone decisions where possible.
	18	* Design for debugging and rework. The CPU should be easily instrumented.
	19	* Defining what IS NOT is as important as defining what IS.
	20	* Don't add functionality unless a real application requires it.
	21	* Don't get bogged down in I/O. Use modern hardware where helpful.
	22	* Worse is better. If 90% functionality can be implemented for 10% effort, do
	23	it and move on.
	24	* The only thing worse than generalizing from a single example is generalizing
	25	from no example at all.