Why not gnat Ada in gcc?

[Robert Dewar]

<<There's always the GNAT for DOS people.
>>

Yes, and that's actually an interesting example. The original DOS port
was done by a volunteer (Doug Rupp, who now works full time for ACT).
It was quite a tricky port, but once done, the work to update it to
new versions of the GNAT sources is quite straightforward. However,
disappointingly the DOS version has not really kept up with current
development, let alone contributed development input of its own.

I think a more promising source of energy is the GNAT/GNU/Linux cooperation
which has for example generated RPM's for each new release of GNAT, and 
usually quite quickly, although the latest 3.13p release has not yet got
RPM's available.

I actually guess that the most likely source of useful contributions to 
GNAT technology will come from new tools, new libraries, and new bindings.
And these are always welcome, and pretty much independent of the continued
development of the front end.

The front ends for modern compilers are remarkably complex, so it is not
surprising that in both the case of GNAT and g++ the number of significant
contributors to the front end will remain small.