Subtype predicates

Subtype predicates

[Anh Vo]

The following type definition is rejected by GNAT-GPL-2016 with error message 'prefix of "First" attribute cannot be the current instance of a scalar type'. 

type Human_Life_Span_Type is range 0 .. 130 
  with Default_Value => Human_Life_Span_Type'First;


Is it a legit syntax error based on 3.2.4 26/3 of LRM? 

Anh Vo

Re: Subtype predicates

[AdaMagica]

Am Dienstag, 23. August 2016 22:14:30 UTC+2 schrieb Anh Vo:
> type Human_Life_Span_Type is range 0 .. 130 
>   with Default_Value => Human_Life_Span_Type'First;

No, I think it's the rule that the name of the entity declared in a declaration (here Human_Life_Span_Type) may not be used again in the same declaration. (You cannot use something that is to be defined recursively in the definition.)

I just cannot find the RM chapter and verse.

Re: Subtype predicates

[Shark8]

On Wednesday, August 24, 2016 at 7:06:10 AM UTC-6, AdaMagica wrote:
> Am Dienstag, 23. August 2016 22:14:30 UTC+2 schrieb Anh Vo:
> > type Human_Life_Span_Type is range 0 .. 130 
> >   with Default_Value => Human_Life_Span_Type'First;
> 
> No, I think it's the rule that the name of the entity declared in a declaration (here Human_Life_Span_Type) may not be used again in the same declaration. (You cannot use something that is to be defined recursively in the definition.)
> 
> I just cannot find the RM chapter and verse.

I think it's different for aspects -- similar that Rosen's Trick the name of the declaration is referent to the instance of the type. So we could say something like:

   -- A string of upper-case letters followed by a single digit.
   Subtype ID_String is new String
   with Dynamic_Predicate => 
     (For all Index in ID_String'Range => 
       (if Index /= ID_String'Last
        then ID_String(Index) in 'A'..'Z'
        else ID_String(Index) in '0'..'9'
       )
     );


And since [some] aspects are essentially attribute-specification clauses (eg "Size => 8" = "For T'Size use 8;") there is some sense in thinking it is/should-be legit.


OTOH, you could certainly consider something like ID_String's predicate as being part of the definition.

Re: Subtype predicates

[Anh Vo]

On Wednesday, August 24, 2016 at 9:45:52 AM UTC-7, Shark8 wrote:
> On Wednesday, August 24, 2016 at 7:06:10 AM UTC-6, AdaMagica wrote:
> > Am Dienstag, 23. August 2016 22:14:30 UTC+2 schrieb Anh Vo:
> > > type Human_Life_Span_Type is range 0 .. 130 
> > >   with Default_Value => Human_Life_Span_Type'First;
> > 
> > No, I think it's the rule that the name of the entity declared in a declaration (here Human_Life_Span_Type) may not be used again in the same declaration. (You cannot use something that is to be defined recursively in the definition.)
> > 
> > I just cannot find the RM chapter and verse.
> 
> I think it's different for aspects -- similar that Rosen's Trick the name of the declaration is referent to the instance of the type. So we could say something like:
> 
>    -- A string of upper-case letters followed by a single digit.
>    Subtype ID_String is new String
>    with Dynamic_Predicate => 
>      (For all Index in ID_String'Range => 
>        (if Index /= ID_String'Last
>         then ID_String(Index) in 'A'..'Z'
>         else ID_String(Index) in '0'..'9'
>        )
>      );
> 
> 
> And since [some] aspects are essentially attribute-specification clauses (eg "Size => 8" = "For T'Size use 8;") there is some sense in thinking it is/should-be legit.

The 3.5 56.3/3 says that the Default_Value shall be static. Thus, S'First did not comply with Static Semantics. The compiler error should have complained about aspect "Default_Value" requires static expression instead of prefix of "First"... This caused the confusion in the first place.

Re: Subtype predicates

[Shark8]

On Wednesday, August 24, 2016 at 3:18:47 PM UTC-6, Anh Vo wrote:
> On Wednesday, August 24, 2016 at 9:45:52 AM UTC-7, Shark8 wrote:
> > On Wednesday, August 24, 2016 at 7:06:10 AM UTC-6, AdaMagica wrote:
> > > Am Dienstag, 23. August 2016 22:14:30 UTC+2 schrieb Anh Vo:
> > > > type Human_Life_Span_Type is range 0 .. 130 
> > > >   with Default_Value => Human_Life_Span_Type'First;
> > > 
> > > No, I think it's the rule that the name of the entity declared in a declaration (here Human_Life_Span_Type) may not be used again in the same declaration. (You cannot use something that is to be defined recursively in the definition.)
> > > 
> > > I just cannot find the RM chapter and verse.
> > 
> > I think it's different for aspects -- similar that Rosen's Trick the name of the declaration is referent to the instance of the type. So we could say something like:
> > 
> >    -- A string of upper-case letters followed by a single digit.
> >    Subtype ID_String is new String
> >    with Dynamic_Predicate => 
> >      (For all Index in ID_String'Range => 
> >        (if Index /= ID_String'Last
> >         then ID_String(Index) in 'A'..'Z'
> >         else ID_String(Index) in '0'..'9'
> >        )
> >      );
> > 
> > 
> > And since [some] aspects are essentially attribute-specification clauses (eg "Size => 8" = "For T'Size use 8;") there is some sense in thinking it is/should-be legit.
> 
> The 3.5 56.3/3 says that the Default_Value shall be static. Thus, S'First did not comply with Static Semantics. The compiler error should have complained about aspect "Default_Value" requires static expression instead of prefix of "First"... This caused the confusion in the first place.

Why wouldn't 'First (on the discrete-type) be considered Static? It's known at compile-time when the compiler chooses the representation.

Re: Subtype predicates

[Jeffrey R. Carter]

On 08/24/2016 06:06 AM, AdaMagica wrote:
> Am Dienstag, 23. August 2016 22:14:30 UTC+2 schrieb Anh Vo:
>> type Human_Life_Span_Type is range 0 .. 130 
>>   with Default_Value => Human_Life_Span_Type'First;
> 
> No, I think it's the rule that the name of the entity declared in a declaration (here Human_Life_Span_Type) may not be used again in the same declaration. (You cannot use something that is to be defined recursively in the definition.)
> 
> I just cannot find the RM chapter and verse.

No, it's because of ARM 13.1.1(12/3): "If the associated declaration is a
type_declaration, within the aspect_definition ... the name of the first subtype
denotes the current instance of the type (see 8.6)." For a declaration:

V : Human_Life_Span_Type;

the current instance is V, so this says V's default value is V'First. Since you
can't apply 'First to a scalar object, the aspect is in error.

-- 
Jeff Carter
"He had no conception of the instrument. He
was blowing into it."
Take the Money and Run
135

Re: Subtype predicates

[Shark8]

> No, it's because of ARM 13.1.1(12/3): "If the associated declaration is a
> type_declaration, within the aspect_definition ... the name of the first subtype
> denotes the current instance of the type (see 8.6)." For a declaration:
> 
> V : Human_Life_Span_Type;
> 
> the current instance is V, so this says V's default value is V'First. Since you
> can't apply 'First to a scalar object, the aspect is in error.

Hm, another argument for a 'Type attribute, then we could say

type Human_Life_Span is range 0 .. 130
  with Default_Value => Human_Life_Span'Type'First; 

Re: Subtype predicates

[Jeffrey R. Carter]

On 08/24/2016 05:08 PM, Shark8 wrote:
> 
> Hm, another argument for a 'Type attribute, then we could say
> 
> type Human_Life_Span is range 0 .. 130
>   with Default_Value => Human_Life_Span'Type'First; 

Perhaps. The ARG seem leary of adding features where there's a work around, and
you can always say

type Human_Life_Span is range 0 .. 130 with
   Default_Value => 0;

If you're worried about Human_Life_Span'First and its default value diverging,
you can instead say

Human_Life_Span_First : constant := 0;

type Human_Life_Span is range Human_Life_Span_First .. 130 with
   Default_Value => Human_Life_Span_First;

-- 
Jeff Carter
"He had no conception of the instrument. He
was blowing into it."
Take the Money and Run
135

Re: Subtype predicates

[Anh Vo]

On Wednesday, August 24, 2016 at 5:31:28 PM UTC-7, Jeffrey R. Carter wrote:
> On 08/24/2016 05:08 PM, Shark8 wrote:
> > 
> > Hm, another argument for a 'Type attribute, then we could say
> > 
> > type Human_Life_Span is range 0 .. 130
> >   with Default_Value => Human_Life_Span'Type'First; 
> 
> Perhaps. The ARG seem leary of adding features where there's a work around, and

Yes, there are a number of ways to work around. Here is my preferred way

Min_Age : constant := 0;
Max_Age : constant := 130;

type Human_Life Span_Type is new Integer range Min_Age .. Max_Age
  with Default_Value => Min_Age;

Anh Vo

Re: Subtype predicates

[Randy Brukardt]

"Shark8" <onewingedshark@gmail.com> wrote in message 
news:8955f6b0-18c6-41b9-91bf-b756189f53db@googlegroups.com...
...
> Hm, another argument for a 'Type attribute, then we could say
>
> type Human_Life_Span is range 0 .. 130
>  with Default_Value => Human_Life_Span'Type'First;

We examined that extensively, since someone (you, perhaps??) already 
suggested it. It's a can of worms, in fact a case of worms. See AI12-0123-1 
for some problems.

The odds of Ada adding this attribute are roughly the same as me marrying a 
supermodel.

                                   Randy.

Re: Subtype predicates

[J-P. Rosen]

Le 31/08/2016 à 20:16, Randy Brukardt a écrit :
> The odds of Ada adding this attribute are roughly the same as me marrying a 
> supermodel.
> 
Well, you often say that you married Ada... Isn't she a supermodel?

-- 
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

Re: Subtype predicates

[Randy Brukardt]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 512 bytes --]


"J-P. Rosen" <rosen@adalog.fr> wrote in message 
news:nq8no1$fen$1@dont-email.me...
> Le 31/08/2016 à 20:16, Randy Brukardt a écrit :
>> The odds of Ada adding this attribute are roughly the same as me marrying 
>> a
>> supermodel.
>
> Well, you often say that you married Ada... Isn't she a supermodel?

<Grin>

She's not good at the physical aspects of marriage, though. (Come to think 
of it, the average supermodel probably isn't, either [away most of the 
time].)

                                Randy.

Re: Subtype predicates

[Shark8]

On Wednesday, August 31, 2016 at 12:16:54 PM UTC-6, Randy Brukardt wrote:
> "Shark8" wrote in message 
> ...
> > Hm, another argument for a 'Type attribute, then we could say
> >
> > type Human_Life_Span is range 0 .. 130
> >  with Default_Value => Human_Life_Span'Type'First;
> 
> We examined that extensively, since someone (you, perhaps??) already 
> suggested it. It's a can of worms, in fact a case of worms. See AI12-0123-1 
> for some problems.
> 
> The odds of Ada adding this attribute are roughly the same as me marrying a 
> supermodel.
> 
>                                    Randy.

In this particular case it could be like the Fn'Result attribute: only available/valid in the function's aspects -- only in this case being the type's.