Re: Multiple reasons for failure of Ariane 5 (was: Re: Ariane 5 - not an exception?)

[Ken Garlington <garlingtonke@lmtas.lmco.com>]

Sandy McPherson wrote:
> 
> Dear Ken and Robin,
> 
> Can't you guys get your own newsgroup, use email, the phone, or get
> married or something. I used to be amused by your sometimes ungrounded
> mutual accusations, however it is wearing a bit thin now. The Ariane 5
> incident will go down Space Industry in history as the most spectacular
> software failure so far. The reasons behind it are manifold, but I can't
> remember the last time this thread mentioned Ariane 5.

Your newsreader (or is it mine?) must be lagging. I've discontinued that thread.

The claim made by Robin in the thread (and then denied, and then re-stated) was
that the Ariane 5 incident would not have happened if PL/I programmers had been
used. I dsagreed with that point. Except for the curious wanderings into the
SQRT world, I've certainly tried to stay on that point.

> I tended to think Robin was taking part in that well known (at least to
> British, Aussies Kiwis etc.) activity of mug-baiting: you make an
> outrageous statement wait for someone to take it seriously and then have
> a good laugh watching them get wound up. However I'm not sure anymore.
> If this is a real discussion, I think it is quite clear that neither of
> you is going to win.

I've stated more that once that I think Robin is just doing this for a laugh,
although I have received some e-mail from others that claims he's completely
serious.

> I don't think there has been one scrap of evidence to support Robin's
> claims for divine superiority of homo PLIs, how can a NEW PL/I
> programmer be any better than a NEW Ada programmer, does he download
> experience from the manual?

I agree. 

> If Robin's point is reduced purely to
> experience in the industry, then the the language become less important
> as the number of years experience increases. Most people don't core dump
> and forget their previous experience when they are required to use a new
> language. Unfortunately there aren't many programmers around with
> 30+years of experience in any (programming?) language, who actually work
> at the "coal face", so we will never know if the 50+ PL/I guys would
> have done a better job than the 50+ Ada guys, but I think we could prove
> that the 50+PL/Iers would do a better job than the usual bunch of 25-30
> year olds who generally do the coding (and whose main ambition is to
> stop coding as soon as possible, take up a management post and be able
> to make the decisions which make the headllines).

That's a fair point, but I would also say that a 50+ programmer who has not been
developing real-time operational flight programs might not be able to beat
a 30-year old who has spent five years in that field. "Experience" has to be
quantified not just in years, but in terms of the type of experience. Even
within the real-time domain, the kind of experience gained in development of
automated banking systems might not be fully transferable to a safety-critical
embedded application, for example.

I knew that PL/I is used in real-time applications, but I didn't think that it
had been used much for embedded real-time safety critical flight applications.
(If someone knows of any recent projects where PL/I has been used in this domain,
I'll gladly be corrected). Ada, on the other hand, does get used in this domain,
both in the military and commercial world. So, if I was building the Ariane 5
inertial reference system, and the only question I could ask was: "Are you more
familiar with Ada or PL/I?", I'd go with the Ada programmer. Of course, in real
life, that's not the most important question.

> 
> I am not in a position to judge if the exception handling in PL/I is
> better than that in Ada, I do object to this statement though:
> 
> ++robin says:
> ---On the contrary, they do, because they have had more experience
> in using the exception capabilities on a daily basis, not just
> in dedicated real-time systems.
> 
> The very first Ada I wrote more than 8 years ago used the exception
> capabilities and I have used it regularly since (but not on a daily
> basis), I have also never used Ada in an embedded system, only on big
> clunky machines with big clunky operating systems. There are certain
> (non)features of Ada which had me pulling my hair out, but these seem to
> have been fixed by Ada95- the exception handling wasn't one of them.
> 
> The only PL/I program I have seen was such a pile of garbage, (it was
> more then 5000 lines long with no indentation no comments and no
> sub-programs), it was an operational analysis tool which happened to
> prove the invincibility of the air force, because the random number
> generator was broken. Not even the cleverest of exception handling
> capabilities would have been able to rescue pilots from the mess this
> thing would have gotten them into. This does not lead me however to the
> conclusion that PL/I is crap. IMO it is complete nonsense to trash Ada
> (or any other language, including PL/I), simply because people make
> stupid management/ design/ coding decisions, because we would quickly
> have no languages left to use!!

I've seen (and, I think, written) good PL/I programs, and have seen (and
probably written) poor Ada programs. I think we're in total agreement on
this point: there's no magic here. I happen to think Ada is well-designed
for what I use it for, but I am painfully aware that programmers, even
experienced programmers, can make mistakes in any language. 

> It takes two to cause an argument, but only one to step out and stop it.
> Would one of you please come to your senses and spare the rest of us.

See my last post to Robin.

> 
> Sandy

-- 
LMTAS - "Our Brand Means Quality"