"J-P. Rosen" <rosen@adalog.fr> wrote in message 
news:logs26$5sl$1@dont-email.me...
> Le 26/06/2014 00:40, Yannick Duchêne (Hibou57) a écrit :
>>> I have similar concerns now they are the ones 'controlling' the future 
>>> is
>>> ASIS, as there's no external agency defining the packages, [.]
>>
>> Isn't it an ISO standard? Or it isn't anymore?
> There is a standard for Ada 95. There was a decision not to make a
> standard for Ada 2005, now that Ada2012 is out.
>
> The ARG decided to standardize existing practice rather than try
> inventing. There should be a standard, but only after it is stabilized
> on the side of the providers (TBH: AdaCore). But AdaCore developments
> are made available to some members of the ARG (!) to check that they are
> acceptable.

Since the vendors were ignoring what the ARG was trying to do, and there is 
very little use of "standard" ASIS anyway (most applications only work with 
one implementation -- we really only need a standard for applications that 
will be ported from one implementation to another), it isn't worth anyone's 
effort to make a standard. Vendors claim that their customers don't care 
about a standard for ASIS. If ASIS customers were to demand that their 
vendors supported a standard, then things would be different, but there is 
no evidence of that (beyond Mr. Rosen -- which is surely not enough).

The ASIS 95 standard is quite a mess; it wouldn't be possible to create an 
ASIS implementation just from reading the standard. One would have to see 
what other implementations do. (That's true to some extent for all 
standards, but the ASIS standard is worse. And, at least for the Ada 
standard we have the ACATS to provide some additional insight into what a 
correct implementation needs to do.)

                              Randy.