GNAT 4.8 bug with subtype predicate

GNAT 4.8 bug with subtype predicate

[Yannick Duchêne (Hibou57)]

Just out of curiosity, tell me what you get if you try this in a package  
specification:


     type T is range 1 .. 100
        with Dynamic_Predicate => T not in 10 .. 15;

     function F (I : T) return Positive
        is (Positive'Val (T'Pos (I)));


Do you get an error from GNAT complaining about “invalid use of subtype  
mark in expression or call” on the line holding the predicate?

That's what I get with GNAT from GCC 4.8.

The error disappears if I change `F` into:


     function F (I : T) return Positive
        is (Positive (I));


But that may not be allowed, if say T is a range, and the function returns  
Character (the direct conversion is not allowed, and requires 'Val/'Pos).

The error also disappears if I only have this in the public part


     function F (I : T) return Positive;


…then have this in the private part:


     function F (I : T) return Positive
        is (Positive'Val (T'Pos (I)));


-- 
“Syntactic sugar causes cancer of the semi-colons.” [1]
“Structured Programming supports the law of the excluded muddle.” [1]
[1]: Epigrams on Programming — Alan J. — P. Yale University

Re: GNAT 4.8 bug with subtype predicate

[Yannick Duchêne (Hibou57)]

Le Thu, 07 Mar 2013 03:11:55 +0100, Yannick Duchêne (Hibou57)  
<yannick_duchene@yahoo.fr> a écrit:
> The error also disappears if I only have this in the public part
>
>
>      function F (I : T) return Positive;
>
>
> …then have this in the private part:
>
>
>      function F (I : T) return Positive
>         is (Positive'Val (T'Pos (I)));

Forget to say that may not be OK neither, if `F` is intended to be defined  
in the public part of the package spec'


-- 
“Syntactic sugar causes cancer of the semi-colons.” [1]
“Structured Programming supports the law of the excluded muddle.” [1]
[1]: Epigrams on Programming — Alan J. — P. Yale University