Ada policy

[Joe Buck]

On Wed, Sep 01, 2004 at 02:19:45PM -0400, Robert Dewar wrote:
> To try to get this a bit back on track .. I would like to ask the question
> of why Zack and perhaps some others think it would be desirable if GNAT 
> could be compiled with other proprietary Ada compilers. I really have not
> been able to understand any reason why this would be desirable.

What bothers me (and what prevents me from doing testing of Ada) is not
that GNAT won't compile with proprietary Ada compilers, but that a
specific release of GNAT will only compile with certain very specific
other releases of GNAT, and that this changes all the time.  That makes it
very hard to build.

If GNAT relies on extensions, and the extensions that it relies on
constantly change, it makes life very difficult for builders and testers.

One way to avoid this, it seems, would be to try to use only standard Ada.
This would permit other Ada compilers to compile GNAT, but that's not the
point; the point is to lessen the bootstrap problem.  You have chosen
otherwise, and I don't know enough to say with any confidence whether this
is a mistake, just that there is a price for it.