Print

Print


> >                  Macro
> >                 Assembly  Ada 83   C ++
> > Source Code
> > Required in LOC   10,000  2,000  1,200
> >
> > Activities      Person-Months of Effort
> >
> > [snip]
> >
> > LOC/Staff Month
> > (Total Project)    417.0  200.0  192.0
> >
> > Percent of Project
> > for Coding and Test  50%   33%   23%
>
> i would expect these figures to be out of date now, given Ada95, and
> more mature C++ compilers that implement a greater part of the (soon to
> be ratified) standard.

Yes, but he specifically said Ada _83_.  His "language level" table, rates
languages for productivity as (this is only a subset of a HUGE table):

        C               2.0
        Fortran 66      2.5
        Fortran 95      4.5
        Ada83           4.5
        FORTH           5.0
        LISP            5.0
        C++             6.0
        Java            6.0
        Ada95           6.5
        APL            10.0
        Eiffel         15.0

Note that "productivity" here basically correlates to the number of SLOC
required to implement one "function point" and says nothing about
robustness, error rates, and maintenance costs.  Also, many of the
languages he lists would get results far different than his predictions if
someone were foolish enough to use them for an application domain outside
of their designed use.  (For example, Excel 5 spreadsheets--the actual
spreadsheet, not any embedded Viual Basic--is rated 57.0.  VB itself rates
7.0 to 11.0 depending on the version.)

I quoted Eiffel for interest, since Eiffel fans have argued with us Ada
fans so much in the past.  I asked SPR for the basis of the Eiffel, C++,
and Ada numbers, but have yet to see an answer.