Re: ANNOUNCE: Ada success story in IEEE Software

["Pat Rogers" <progers@classwide.com>]

"Jeffrey Carter" <jrcarter@acm.org> wrote in message
news:3C3CB31E.BC6FDBED@acm.org...
> Ted Dennison wrote:
> >
> > In article <3c3aea7b$1@pull.gecm.com>, Martin Dowie says...
> > >
> > >"Dale Stanbrough" <dstanbro@bigpond.net.au> wrote in message
> > >news:dstanbro-FFA1D2.23113708012002@mec2.bigpond.net.au...
> > >> Rod Chapman wrote:
> > >>
> > >> > factor in the success of the project.  The final system achieved
> > >> > 0.04 defects per kloc (that's 4 defects in 100,000 lines of code) post-
> > >> > delivery, which compares favourably with industry norms.
> > >>
> > >> I'm not sure I follow this. If this result is the industry norm,
> > ..
> > >He said it "compares favourably with industry norms" not "matches the
> > >industry norms"
> >
> > I suspect that may have been a bit of that famous British understatement
too. I
> > think our last (non safety-critical) Ada project would have had a defect
count
> > orders of magnitude higher than that.
>
> I recall seeing the "industry norm" given as 1 defect per kloc.

In a study described in a paper published in 1986, Herbert and Myron Hecht found
that for each million lines of code, 20,000 bugs existed.  Normally 90% would be
found by testing.  Another 200 would be found during the first year of operation
by users, leaving 18,000 undetected bugs.  Regular maintenance would fix 200
bugs, but also introduce 200 new ones.

Things have probably progressed *a little* since then, but not much; certainly
not orders of magnitude.  Too, that's just one study (of several large systems),
but the magnitude is astounding.  (Both individuals are extremely well respected
in the software fault tolerance field.)

The paper is:
H. Hecht and M. Hecht, "Software Reliability In The Systems Context," IEEE
Transactions On Software Engineering, vol. 12, no. 1, pp. 51-58, 1986.

---
Patrick Rogers                       Consulting and Training in:
http://www.classwide.com          Real-Time/OO Languages
progers@classwide.com               Hard Deadline Schedulability Analysis
(281)648-3165                                 Software Fault Tolerance