> This reminds me of something else that may be useful, which is the Bell Labs
> Technical Journal article describing the programming of some switching system
> they developed (some of you may remember the one, it was mentioned some time ago
> on c.l.a). If I remember, I'll try to get the URL for it, but it was basically a
> description of how the team had gone about improving the reliability of coding
> in C. Essentially they came up with around 10 points or something related to use
> of C that should be either avoided or checked specifically to make sure they
> were doen correctly. Interestingly the majority of the points were ones that
> would have been picked up by an Ada compiler!
> It was quite interesting and if anyone know of it, please feel free to post the
> URL. Otherwise I'll try to post it tomorrow.
There is a "new" link to this report on (see if I get it right this time):
By the way, look at
The last item, "Published Results..." doesn't mention directly three things that
I deduced (perhaps wrongly) from reading between the lines:
1. The cost differences were measured in terms of "first month," second
month," etc. after deployment. But the Ada deployment was four
years later than the "mix" deployment. I presume salaries were
higher, and... is there inflation in Malaysian money?
2. The "mix" version had the benefit of at least two years experience
maintaining an earlier version written in the same languages. The
Ada version was a new implementation (although I'm sure there must
have been at least a little knowledge carry-over).
3. The programmers on the "mix" version had several years experience in
the languages used. The programmers for the Ada version (possibly
the same people) learned Ada in order to do that version.