Re: Why not gnat Ada in gcc? (really ia64 port and NDAs)

[Joe Buck <jbuck at racerx dot synopsys dot com>]

> Technically, there is a conflict between NDAs and the GPL.  As a
> practical matter, it is a moot issue.  This is because all parties
> involved agree voluntarily and informally (i.e. a gentlemen's agreement)
> to respect the terms of both the NDA and the GPL despite the conflict.
> This makes it possible to work on GPL software under NDA.

Furthermore, in the particular case of the ia64 port, RMS was informed
about the plans and accepted them, at least in outline, and progress
reports were regularly made to the public (on the ia64linux web site among
other places).  As the FSF is the only organization with standing to sue
if the GPL is violated, and RMS makes decisions for FSF, that's pretty
much the end of the legality argument.

But we're entitled to ask people for more than just the bare minimum
required to escape prosecution, lawsuits, or jail.  We're supposed to be a
community here.  The ia64 project has been reasonably good at
communication, at least for those who knew where to look, but some other
gcc-related projects have been done much more quietly and privately than
many of us would like.

The document http://gcc.gnu.org/contributewhy.html is highly relevant
here.  (The text is mostly by Jeff Law, RMS, and me, though the whole SC
reviewed it, so I'd like to think of it as something both the FSF and the
commercial contributors to gcc can sign up to).  It acknowledges that
sometimes there are reasons for a time-limited NDA, but still strongly
requests that other developers be kept informed.  In my mind, this
document is making the same requests of Red Hat, CodeSourcery, and ACT as
it is making of "outside developers".

In the case of the IA64 port, it would have been nice if more progress
reports had been made available to the gcc list, this could have been done
without violating anyone's trade secrets.  But generally speaking, I don't
have a problem with that port in particular.  I'd like to see more
discussion of some of the other major projects that Red Hat has taken on
-- I accept that the customer who pays gets first access to the code,
but more input from non-Red Hat developers on the design would probably
mean that both Red Hat's customers and the general public winds up with
a better solution.