Re: Introductory Ada Programming Book

["J-P. Rosen" <rosen@adalog.fr>]

Le 04/01/2017 à 14:44, raph.amiard@gmail.com a écrit :
> I'm in contact with the author so I'll pass any fixes up to him if you don't mind
Well, I suggested help from the community at large because currently I
don't have the time to make an extensive review. Otherwise, I'd write to
him myself...

Some have already responded; my $0.02

>> Oddly, variables are not initialised by default
>> (Personal opinion: default initialization is a huge mistake)
> 
> Ok but, besides the "oddly" qualifier, his statement is correct, right ?
> 
Asking for automatic initialization comes generally from people using
the C like languages, who automatically think of 0 as the initial value.
A truely magical value, meaning number 0, false, string termination,
etc. It makes no sense for Ada, where types may not include 0 as an
allowed value. You can even have types with no value at all; what would
the initial value be?

But the issue is worth than that. A program that does not initialize a
variable before using it has a bug. If the program works because the
compiler provided a default value that happened to have an appropriate
value, it's a buggy program that works by chance. In the safety critical
world, we don't want programs that work by chance; we want programs
without bugs. That's why the only acceptable initial value is an invalid
value (as provided by Initialize_Scalars), because it maximizes the
probability of catching the bug early.

>> Ada is, alas, a one-pass language, so things need to be prototyped if
>> you're going to refer to them before they're defined.
> 
> This is generally true, and the areas of the language where it's not
(aspects) are full of gotchas. While it's not always true, my feeling is
that there *was* a strong push to keep Ada compilable by a one pass
compiler (which GNAT is, in broad strokes).
> 
I don't think that one pass was ever a requirement for Ada. Claiming
that some things are "unfortunately" the way they are to make compilers
naively simpler sends a wrong message, and is unfair to the huge work
performed by compilers.


-- 
J-P. Rosen
Adalog
2 rue du Docteur Lombard, 92441 Issy-les-Moulineaux CEDEX
Tel: +33 1 45 29 21 52, Fax: +33 1 45 29 25 00
http://www.adalog.fr