Prev: Slightly OT: Debian Community Poll
Next: A Newbie to the group.
From: Adam Beneschan on 14 Jun 2010 20:58
On Jun 14, 5:01 pm, "Peter C. Chapin" <pcc482...(a)gmail.com> wrote:
> Yannick Duchêne (Hibou57) wrote:
> > What disturbed me, is that a literal here, match an ASTERIX while it is  
> > statically out-of range ?
>
> I think that's accepted because the literal is a Universal Integer.

No, that's not the reason. In fact, with one compiler, the above code
gives me a warning, but if I change the out-of-range upper bound by
sticking another zero on it:

subtype IDEFIX is ASTERIX range 1..200_000;

then it gives me an error and rejects the program---as I expected.

Here's the reason: The only error that has to be caught at compile
time is when the value is out of the *base* *range* of the expected
type (see 4.9(35)). The base range for ASTERIX is not -5000 .. 10000;
it's some other range selected by the compiler, and I think in this
case this compiler probably selects
-32768 .. 32767.

I can understand why this all might be confusing to a newcomer. I'm
not exactly sure why the language rules were done this way. But I do
know that good practice often means writing programs in such a way
that you should be able to change a program's behavior by changing a
constant without having to go through the whole rest of the code
figuring out what else might have to be changed (sorry, I forgot the
term for that), and that means that sometimes a program will have IF
statements where the condition will always be false (or true), and you
don't want to have language rules that force a program to be rejected
if it contains a statement that is statically known to fail inside a
block of code that will never be executed. Maybe that was the
motivation.

-- Adam
From: Nobody on 14 Jun 2010 21:05
On Tue, 15 Jun 2010 01:31:47 +0200 Yannick Duchêne (Hibou57) wrote:

> [ARM 2005 3.5(5)] just says:
>> For a subtype_indication containing a range_constraint, either directly
>> or as part of some other scalar_constraint, the type of the range shall
>> resolve to that of the type determined by the subtype_mark of the
>> subtype_indication.
>
The annotated ARM says:

3.5 Scalar Types, 8.a
Ramification: Only range_constraints (explicit or implicit) impose
conditions on the values of a scalar subtype. The other
scalar_constraints, digit_constraints and delta_constraints impose
conditions on the subtype denoted by the subtype_mark in a
subtype_indication, but don't impose a condition on the values of the
subtype being defined. Therefore, a scalar subtype is not called
constrained if all that applies to it is a digits_constraint. Decimal
subtypes are subtle, because a digits_constraint without a
range_constraint nevertheless includes an implicit range_constraint.

What I understand is that a subtype shall in maximum contain all values
of the subtype indicator, i.e. the superordinated type. Therefore the
definition

type ASTERIX range +5_000..10_000;
subtype IDEFIX is ASTERIX range 1..20_000;

should in ANY case cause a compile time error!!!

At least I would expect this behaviour.

Regards

Nobody


From: Adam Beneschan on 14 Jun 2010 21:20
On Jun 14, 6:05 pm, Nobody <nob...(a)nowhere.org> wrote:
> On Tue, 15 Jun 2010 01:31:47 +0200 Yannick Duchêne (Hibou57) wrote:
>
> > [ARM 2005 3.5(5)] just says:
> >> For a subtype_indication containing a range_constraint, either directly
> >> or as part of some other scalar_constraint, the type of the range shall
> >> resolve to that of the type determined by the subtype_mark of the
> >> subtype_indication.
>
> The annotated ARM says:
>
> 3.5 Scalar Types, 8.a
> Ramification: Only range_constraints (explicit or implicit) impose
> conditions on the values of a scalar subtype. The other
> scalar_constraints, digit_constraints and delta_constraints impose
> conditions on the subtype denoted by the subtype_mark in a
> subtype_indication, but don't impose a condition on the values of the
> subtype being defined. Therefore, a scalar subtype is not called
> constrained if all that applies to it is a digits_constraint. Decimal
> subtypes are subtle, because a digits_constraint without a
> range_constraint nevertheless includes an implicit range_constraint.
>
> What I understand is that a subtype shall in maximum contain all values
> of the subtype indicator, i.e. the superordinated type. Therefore the
> definition
>
> type ASTERIX range +5_000..10_000;
> subtype IDEFIX is ASTERIX range 1..20_000;
>
> should in ANY case cause a compile time error!!!

3.5(8), which you're reading, is in a "Dynamic Semantics" section.
Whatever you see there explains how the program will behave at run
time. Nothing in a Dynamic Semantics section should ever affect
whether you get an error at compile time or not, I think. The rules
that cause compile-time errors are in "Legality Rules" sections, and
perhaps in "Static Semantics".

As for why this would cause an error at run time but not at compile
time, I've tried to explain already why it might be a good idea to
have the language rules this way. What this means is that if your
compiler tells you that something you wrote will raise an exception at
run time, take it seriously, as seriously as you'd take an actual
error.

-- Adam

From: Jeffrey R. Carter on 14 Jun 2010 16:36
Nobody wrote:
>
> I am new to Ada and found the following declaration in my text book:
>
> type ASTERIX is range -5_000..+10_000;
> subtype IDEFIX is range 1..20_000;

Which text book are you using? All caps for identifiers has not been commonly
used for 15 years or more.

If you really found that in your text, you should get another text. The
declaration of Idefix is invalid, as it does not give the parent subtype. A
valid subtype declaration is

subtype Positive is Integer range 1 .. Integer'Last;

You are correct that the range of a subtype cannot exceed that of its parent.

--
Jeff Carter
"Apart from the sanitation, the medicine, education, wine,
public order, irrigation, roads, the fresh water system,
and public health, what have the Romans ever done for us?"
Monty Python's Life of Brian
80
From: Nobody on 17 Jun 2010 14:54
On Mon, 14 Jun 2010 13:36:13 -0700 Jeffrey R. Carter wrote:

Hi Jeff,

thanks for your answer.

> Which text book are you using? All caps for identifiers has not been
> commonly used for 15 years or more.

Can you give me direction where I can find documents about the current
common use of Ada as you stated above?

Best regards

Nobody



First  |  Prev  |  Next  |  Last
Pages: 1 2 3
Prev: Slightly OT: Debian Community Poll
Next: A Newbie to the group.