Re: what DOES the GPL really say?

[kenner@lab.ultra.nyu.edu (Richard Kenner)]

In article <m2zpq0h7uc.fsf@devo.ridgecrest.ca.us> Ronald Cole <ronald@devo.ridgecrest.ca.us> writes:
>should Dewar consider basing the next public release of GNAT upon an egcs
>snapshot? 

I think you seriously misunderstand the purpose of the EGCS project,
which is to provide a framework within which highly experimental (and
known to be broken) patches to GCC can be developed and perfected.
Those patches are then fed into the normal GCC development process for
further testing and then become parts of GCC releases.

Nobody would want to build a public release of anything based on
software that, by its very design, has serious problems.  That's not
the purpose of this work.

>(It really would be nice to be able to build gcc, g++, f77, and GNAT from
>a unified source tree without having to apply two differing patch
>files and hoping for the best...)

As of the last (or next, I'm not sure of the precise schedule here)
g77 release, the g77 patch set includes all of the GNAT patches.
There is some question about whether the g77 patch set contains
patches that will break GNAT, but none of these patches are actually
required to build a correctly-functioning g77; they are to add
optimizations that are deemed important for Fortran users.  Some of
these optimizations are currently also being experimented with in the
EGCS project.

As far as we know, GCC 2.7.2 plus the patches supplied by GNAT, will
build a functioning g77, g++, and GNAT, with one exception, which relates
to a change in the front-end interface and is what either just has or
soon wil be addressed in the g77 release.