Re: Uninitialized out parameters.

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On 08/04/2016 01:02, Randy Brukardt wrote:
> "AdaMagica" <christ-usch.grein@t-online.de> wrote in message
> news:51041ba8-3243-46bc-b97d-b106d4b357e2@googlegroups.com...
>> Am Donnerstag, 7. April 2016 11:57:02 UTC+2 schrieb
>> Stefan...@uni-weimar.de:
>>
>>>> (2) The latter can be written in current Ada 2012: (that is, with TC1
>>>> implemented)
>>>>   function Y return T is (raise Program_Error);
>>>
>>> Great! I wasn't aware of that. I always thought, the expression in
>>> brackets had to be of type T.
>>>
>>> Stefan
>>
>> A raise_expression matches any type. See RM.
>
> Right. One can think of it sort of like "bottom" in a functional language -
> it represents the "not a value" of any type. (And touch it and you die.)
>
> OTOH, like me, you may not want to think of functional languages much. ;-)

Yes, better think it typed!

It is same as an IEEE float which values are real U {NaN, +Inf, -Inf}. 
Propagating exceptions is a sort of value.

In

    function Y return T is (raise Program_Error);

we assume that T consists of values of T values and exceptions of all 
sorts. That is weakly typed, a bit. But it also a way how to spell Y's 
post-condition: "T or Program_Error raised" could be a type Y returns. 
That would make an implementation (raise Constraint_Error) illegal.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de