Re: Ada News Brief

[Sandy McPherson <sandy@wgs.estec.esa.nl>]

Steve Jones - JON wrote:
> 
> Sandy McPherson wrote:
> >
[snip...]

> > Why are the FAA moving to other languages? In Europe Ada is ATC language
> > number one. Did the FAA get burnt by an Ada development which went out
> > of control. Could someone supply me with references please?
> >
> 
> Well I've worked on the NERC ATC project (The new posh UK system) and am
> now working at Eurocontrol (A conglomeration of all of Europes Aviation
> Authorities to create a unified approach).
> 
> NERC used C to do the GUI side, mainly because at the start there were
> no offical X/Motif bindings.  Eurocontrol is at present using Ada in
> totallity for its simulations (it has the bindings) but will be moving
> more towards C++ as it moves to client server.  The problem is that
> new technoligies come out first in C (now in C++) and projects have to
> go that way.
> 

On a project I worked on a few years ago, we used C for GUIs, precisely
because of the binding problem. Even if you have the bindings, the data
types used by X11 etc. are decidely Ada unfriendly. We limited the C
code to control of the GUI and the putting/getting of data into/from the
widgets. The rest was written in Ada

> As for why the old FAA project failed ?  From talking to people who
> worked
> on it (ie not managers or press releases). There were several horrific
> design decisions made at early stages and an overestimation of just
> what was physically possible and simple.
> 
> If you have a bad design no language would be better than another.
>

Aha, so Ada is the scapegoat?. Were the daft decisions in any way
related to a misunderstanding of the concepts present in Ada though?
 
> On the NERC side though about 90% of the bugs were in the C code of
> which only about half would have been picked up if it had used
> Ada.  The problem was also that GUIs tend to be very complex, although
> no end-user will belive you, and you would expect more problems there.
>

Did you use a good static analyser like QAC, and run time system like
Purify during your testing, these should identify most of the errors
that the Ada compiler and Runtime checks would find. What was the ratio
of Ada/C?. I can imagine that if vanilla C without lint were used, that
this would be the case, but with good support tools, this would seem
rather outrageous. Does this fact not worry you when you think about the
move to C++? Or, is C++ only for the simulators? If so this might be
acceptable, but the thought of a client server ATC system written in C++
gives me the creeps. 

I agree about GUIs, easy to use does not mean easy to build, usually the
contrary.
 
> Ada was absolutely essential however in making sure communications
> between
> systems was correct, I dread to think what would have happened
> with 100 people working in C with interfaces with no size constaints.
> 
C++ isn't any better.

-- 
Sandy McPherson	MBCS CEng.	tel: 	+31 71 565 4288 (w)
ESTEC/WAS
P.O. Box 299
NL-2200AG Noordwijk