Ada policy

[Robert Dewar]

Florian Weimer wrote:

> * Robert Dewar:
> 
> 
>>That's just not true. GNAT has been able to be built from 3.13 for
>>many years (3.13 was released several years ago). We only introduce
>>incompatibilities if we really need to for some significant cleanup
>>or kludge removal. Earlier on that was more common, but we have not
>>needed to do that for many years now.
> 
> 
> I wonder why it's not possible to disable some aspects of -gnatg when
> building the stage1 compiler.  If there was such an option, you might
> be able to do some cleanups (like instrumenting the code with more
> precise pragmas to control warnings) without requiring newer compiler
> versions to bootstrap.

We can consider that sort of thing if we run into it, but as I noted,
we can now bootstrap with very old versions of the compiler (3.13 was
released in the middle of 2000, so over four years ago). I don't think
there is any point at this stage in trying to make it work with older
versions, and at the same time, we don't see any problems in maintaining
the compatibility with 3.13 at least for now. If problems come up we
can discuss them then.

Meanwhile, I really think this is a non-issue. No one has ever had
trouble from trying to use a too-old version of GNAT to bootstrap
that we are aware of.