Re: Ada noob here! Is Ada widely used?

["Yannick Duchêne (Hibou57)" <yannick_duchene@yahoo.fr>]

Le Fri, 04 Jun 2010 14:23:07 +0200, Dmitry A. Kazakov  
<mailbox@dmitry-kazakov.de> a écrit:

> If you mean mathematical statistics, then its applicability depends on
> strict conditions. Prior these established the statistics (samples) of
> failures is just a collection of anecdotes...
As usual, with mathematics and logic, interpretation is an issue. The  
interpretation here should be correlation (I am not arguing this is true  
that software is less or more reliable than mechanical parts, as I don't  
have needed materials to assert anything about it). Statistics are  
intermediate results, and intermediate result are not always  
interpretable, ok, you're right.

> If you mean "lies, damned lies, and statistics" then yes. (Did you know
> that 90% of people died in car accidents had eaten cucumbers shortly  
> before
> the accident? (:-))
This is not even an implication, so it is unlikely this will legitimately  
argue for a correlation. Here is why: “Had eaten cucumbers” may be an  
antecedent of many other things, so this would not be a meaningful  
correlation, and moreover “Had eaten cucumbers” may be an antecedent of  
“accident occurred” as much as “no accident occurred at all”, so this does  
not justify an implication or correlation. Well, I am relying on an  
implicit here, because what is exactly missing, in your example, would  
exactly be statistics about “Had eaten cucumbers” when “no accident  
occurred at all”. Conclusion : one statistic is not relevant alone, it  
needs others, forming a good coverage of different cases (logic needs its  
food).

I was wrong just saying “statistics”, so the reformulation I suggest is  
“this does not invalidate any noticed correlation” (statistics being just  
a tool there, to help see correlation, and multiple statistics are needed  
for various configurations).

I suppose I understand what you mean and was just wrong with my wordings.

Cheers

-- 
There is even better than a pragma Assert: a SPARK --# check.
--# check C and WhoKnowWhat and YouKnowWho;
--# assert Ada;
--  i.e. forget about previous premises which leads to conclusion
--  and start with new conclusion as premise.