> > can you say more about the licensing problems? ...
>
> Well, almost all the M3 code is under less restrictive license than the
> GPL, which allows you to do more or less anything you want with it
> (including commercial use). Only if you pass back changes to the
> copyright owners these will be free to do with them what they like.
right. i love that licensing model and i use it all the time. the
success
of FreeBSD, NetBSD, X11, Apache, BIND, MIT Kerberos, UDel NTP, etc, are
proof
that this licensing model is viable.
> The problem with the gcc backend was that Digital SRC specified a small
> file-based interface which allowed them to put only some lines of C code
> under the GPL and avoid the viral license for the rest of the M3 code,
> which wasn't very well received by the FSF people, who rejected to
> include the backend in their distribution :-(
"viral" is the part i don't understand. if you want some source files to
be
****pped as part of GCC, then those files will have to be licensed under
GPL.
but, the viral nature of GPL only taints the code that ****ps with GCC, not
the rest of the CM3 code base (which will not ****p with GCC or via FSF.)
so while i can wish that the GCC could include some files with BSD-style
licenses, i don't see how the GPL virus affects CM3 at all. the files
you'd
be "giving up" (allowing to be published with a GPL) would still be yours
to
publish under some different license if you wanted to. but there's no
reason
to want to do that, since they wouldn't be useful outside of GCC. if
someone
else improves the files you "give up" to GCC in this way, then their mods
would likely also be available only-under-GPL. those files are just lost
to
you, tainted by GPL, etc. but the rest of CM3, which depends only on a
GCC
that has those files inside it but does not link to them or include text
from
them, would not be tainted.
so, please say more?
> > and speaking of wide use, what does CM3 have or
> > lack with respect to John Polstra's Ezm3?
>
> ... John Polstra actually took a PM3 distribution and removed everything
> from it he didn't need for CVSup (including advanced garbage
collection).
> I don't think EZM3 is the way to go (though it's a great distribution
for
> compiling CVSup).
so, cvsup will build just fine on CM3, and we have no worries about it
depending on some special feature of PM3 or EZM3. that's good, since it
preserves the possibility of CM3 becoming the standard for M3.
> > what's involved, beyond getting past the licensing issues?
>
> Technically, the code could be added very easily to the gcc distribution
> I think. It would not contain a complete M3 compiler then, but just an
M3
> code generator. Aside from the licensing problems we would probably have
> to add some test framework so it can be maintained independently of the
> complete CM3 distribution.
that's all very easy compared to the licensing problems if any. good.
--
Paul Vixie
|
