Path: utzoo!mnetor!tmsoft!dptcdc!lethe!dave
From: dave@lethe.UUCP (Dave Collier-Brown)
Newsgroups: comp.software-eng
Subject: Re: Surface area for software
Summary: one possible formula...
Message-ID: <2481@lethe.UUCP>
Date: 8 May 89 00:18:05 GMT
References: <8026@thorin.cs.unc.edu> <39400020@m.cs.uiuc.edu>
Reply-To: dave@lethe.UUCP (Dave Collier-Brown)
Organization: Officially bimodal now, News courtesy of Systems Software
Lines: 67



Written  5:06 pm  May  3, 1989 by shan@unc.cs.unc.edu:
| Cox in his book, "Object Oriented Programming" p. 16, defines the "surface
| area" as "the number of things that must be understood and properly dealt
| with for one programmer's code to function correctly in combination with
| another's."

In article <39400020@m.cs.uiuc.edu> render@m.cs.uiuc.edu writes:
| I don't think Cox chose the term well (if indeed he did choose the
| term) given what he intends for it to mean. 

  Ok, lets try a few possible meanings and see if we can come up with
something thats usefull:
  let ops be the number of operations defined in a module
      args be the number of parameters and external variables needed
	by a given op.
      unique args be the number of non-duplicate args defined for two or
	more ops.
      common args be the args used by every op.
      let minset be the smallest set of operations which are required
	to do any meaningfull operation via the interface [note 1]

  surface area ::= width * depth
  depth ::= (minset * 1) + (ops * FUDGE1) [note 2]
  width ::=  (common args * 1) + (unique args - common args) * FUDGE2

  As you can see, this is a measure of the length of the parameter lists
(expressed in tokens) times the number of parameter lists.  The required
information is weighted higher than information which is only used in some 
cases... And if you could write parameter-lists succinctly in a table, it
might even be an analogy of the physical surface area of the table. [note 3]

  Now, is it usefull? Is it intuitive?   What do we have to add to it to make
it usefull for something.  Indeed, even if we get it "right", what are we
masuring and why is it important?

[note 1] I don't have a formalism for this, but for the case of file i/o
	the minset would probably be {open, read|write, close}.
[note 2] I'm not even going to **try** to estimate weightings.
[note 3]  An example:

name	returns	args
====	=======	======================
open	fd	name	mode	cookie
close	rc	fd
read	rc	fd	data	size
write	rc	fd	data	size
seek	rc	fg	amount	from

ops = 5
minset = 3
unique args = 9
common args (for minset?) 3 ?

  depth = (3) + (5 * FUDGE1) 
  width   (3) + (9 - 3) * FUDGE2
  surface area = (3 + 5*F1) * (3 + 6*F2)

if both fudges are .1, we get 3.5 * 3.6 or 12.6

--dave
-- 
David Collier-Brown,  | {toronto area...}lethe!dave
72 Abitibi Ave.,      |  Joyce C-B:
Willowdale, Ontario,  |     He's so smart he's dumb.
CANADA. 223-8968      |