larry@JPL-VLSI.ARPA (07/09/86)
Halstead metrics are discussed by Neal Coulter in the March '83 issue of
the IEEE Transactions on SW Engineering ("SW Science & Cognitive Psych,"
p. 166. Also (p. 155, SW Science Revisited: a Critical Analysis of the
Theory and its Empirical Support") by Vincent Shen, Samuel Conte, and
H. E. Dunsmore. I suggest that order of reading because the Coulter
article was easier for me to read.
Elliott Soloway of Yale University (who's doing studies of the actual as
opposed to theoretical behavior of programmers) characterizes the Halstead
metrics as "shallow structure" measures and suggests "deep structure" measures
as being more relevant to programmers and managers. He calls these measures
(I mean structures, sorry) "plans." An example he gives is
while not EOF
read X;
...
end while;
or "iteration protected by a guard."
The two articles also criticize the Halstead measures, and succeeding issues
of ToSWE includes additional discussion.
[Got to go. I'll have to send the "Controversial Position" promised in the
Subject heading some other time.]
Larry @ jpl-vlsi