[SystemSafety] Collected stopgap measures (Hoare)

[Derek M Jones]

Nick,

> started with 18,000 SLoC in 2001/2 and showed the manual code was well
> above 'world class'....but still found some issues.  This took one year
> elapsed time and 6 man/years.  The techniques were refined and used on

18k in 6-man years sounds believable.

> various iterations, eventually doing all 350,000 SLoC in 4 man months, 2
> months elapsed, one month of which was writing the report.  The problem

This sounds completely unbelievable.  Or is a marketing definition of
formal proof being used here?

Was the code automatically generated from a specification containing
2-orders of magnitude fewer lines?

> There used to be some papers on this, but I now longer have access to
> them.... :-(

I can try and track them down, if you have the titles.

-- 
Derek M. Jones           Software analysis
tel: +44 (0)1252 520667  blog:shape-of-code.coding-guidelines.com