Subj : Re: curve for verbosity in a language
To   : comp.programming
From : pantagruel
Date : Sun Aug 07 2005 03:17 am

Well, a propos your point, this is basically why I argued for a
whitespace normalized character count. The salient point though is not
to achieve a perfect metric, but rather to achieve a metric the
shortcomings of which are well understood, so that one could analyze
code in one language and then have a list of characteristics of  that
particular language that would affect the analysis. As such the LOC
metric although imperfect is also quicker, and its drawbacks are
reasonably well understood.

Perhaps however true verbosity is a combination of tokens and average
lines usage. Thus analyse programs written to the problem for tokens,
analyse a dataset of 'average' programs in the languages for LOC.
The results of token analysis is weighted higher but the graph of LOC
can also be significant.
Obviously LOC becomes more pertinent as we get into programs that run
over many pages, especially if, as is often the case with J or K, the
competing programs in those languages can be read as a couple pages of
printout.
As noted by Graham:

"total effort = effort per line x number of lines"


The traditional layout of LOC for a language may exist because of
legibility problems with the language without its use.


This has bearing on the problem because verbosity in a language is
often directly related to comprehensibility of any segment of code.

Graham suggests that the comprehension of code in its totality is more
at issue but if any segment is nigh incomprehensible then how
comprehensible can the total be?

.