Re: GNAT GPL Edition Maintenance and Upgrades

[Jeff Creem <jcreem@yahoo.com>]

Maciej Sobczak wrote:
> Georg Bauhaus wrote:
> 
>> The C++ compiler also has a history of issues... not stopping
>> a release.
>> Why is it considered good enough in spite of the issues?
>> I guess that's the meaning of "good enough"?
> 
> 
> I guess that for any product there exists an arbitrary meaning of "good 
> enough" that can prevent it from being released, ever.
> 
> The version of GNAT that I got with my Linux distribution allows me to 
> overflow the Integer type without any exceptions at run time (and I can 
> also define my own range and overflow it without error). I guess that 
> with some chosen meaning of "good enough" this compiler would not be 
> released.
> 
> 

So just to be totally clear about what other are saying here... Every 
version of GNAT from every source has always acted this way.

It also is non-standard in terms of the way it does elaboration checking 
(but better).

It also is non-standard in terms of the way it does stack checking.

This is not an issue unique to the FSF tree. These are design decisions 
made by Ada Core Technologies many many years ago.