Conference turris::c_plus_plus

3438.0. "Any relationship between the number of code statements and program text size" by UNXA::DERZINSKI () Fri Feb 07 1997 18:54

Question: Is there any data/metrics regarding the c/c++
compilers, showing the relationship between a programs
text size to the number of source code statements. I
realize there are many variables that must be taken
into consideration. However if one took a sample of 100
binaries all linked shared, could I expect a normal
bell curve or distribution?

Here is the reason I am asking this question:  I am
taking a  class on software project planning. Many of
the models used in software project estimation require
that you estimate the numbers of source lines of code.
If you have access to historical data or to projects
with similar sources, no problem.

What if this is not available?

Lets say that you have the binary "foot print" of your
competitors product and your product will be similar in
features and functionality. Assuming they are built
with the same compiler, one can certainly reason that
the bigger the text area, the larger the number of
source lines of code.  Knowing the mean and the
standard deviation of the text-to-statement
relationship one could come up with a range at least.

I could take a few samples to test this, but before
doing so I wanted to see if any data or similar studies
were available.

Any wild guesses?

For example: On average 1 line of code = 100-150 bytes of text

Thanks in advance.

John 

fixed by v6.0