Re: GNAT and GCC 3.0

[dewar@gnat.com (Robert Dewar)]

britt@adapower.net (Britt Snodgrass) wrote in message news:<36c6f8dd.0110051506.3fdabae0@posting.google.com>...
> dewar@gnat.com (Robert Dewar) wrote in message news:<5ee5b646.0110041038.5c321817@posting.google.com>...

> Will having GNAT available as an (optional) part of the 
> GCC source distrubution

    GNAT *is* part of the GCC source distribution. What
    parts you pick up if you do not pick up the whole
    distribution is up to you.

> result is a significantly simpler process for building
> cross-compilers to target additional processor families?    
> I'm thinking of processors such as the AMD 29050 and 
> StrongARM that have long been supported by the GCC C 
> compiler but not by the public versions of GNAT.
> 
> Britt Snodgrass

No. The process is straightforward from a build point of
view with either version of GNAT. Of course the issues of
supporting tasking, and the full tool set are completely
independent of the build process.

Why would you think otherwise?

Porting GNAT to a new target is never a trivial task. We
usually consider that internally a straightforward Unix
port is 6-12 person months with all the necessary tools
and testing, and cross ports, or ports to unusual operating
systems can be an order of magnitude more.

The one thing that having a version of GNAT that is compatible with
GCC 3 helps with is that there are some
new targets that are only available with GCC 3 (and were
not available with GCC 2.8.1), and furthermore some other
ports are in much better shape. But that has to do with
the version of GNAT you are using, not with whether it
is part of the GCC source distribution.

Porting GNAT to other than vanilla Unix systems on
traditional byte addressable machines is definitely
not straightforward. As far as we know the only
such port ever achieved outside ACT was Doug Rupp's
port to DOS (and Doug now works for ACT -- where he is
the primary maintainer of the port of GNAT for VMS -
which is certainly the trickiest of all the ports :-)

I certainly do not want to discourage people from trying,
but on the other hand, it is important to understand the
magnitude of the task at least a little bit!


Robert Dewar
Ada Core Technologies