Dieter Baron <dillo@danbala.tuwien.ac.at> writes:

> On Tue, May 16, 2006 at 09:13:29AM -0400, Greg Troxel wrote:
>   What about HONOUR_LICENSES, which defaults to YES, and
> DISTRIBUTABLE_PACKAGES, which defaults to NO and if set won't build
> packages with NO_BIN_ON_FTP set?  The latter could be set for the
> official bulk builds on slower architectures to trade less error
> detection for faster builds.

I prefer to avoid the word HONOR_LICENSES, since it implies that we
don't sometimes, and the entire point of the larger exercise is to be
careful about licenses.

I think what all we really need is a way to say "build a package even
if LICENSE= is set as long as NO_BIN_ON_FTP is unset" that people
doing bulk builds can choose to use.  So how about
ACCEPTABLE_BIN_ON_FTP, defaulting to NO, and if YES, then bsd.pkg.mk
will set _ACCEPTABLE if NO_BIN_ON_FTP is unset?  This avoids semantic
overload for other things, and solves the current problem with just a
few lines.

>   Then why isn't setting LICENSE, NO_{SRC,BIN}_ON_CDROM, and, for non
> open-source OSs, NO_BIN_ON_FTP enough?  Why do we need
> ONLY_FOR_PLATFORM at all?

We need ONLY_FOR_PLATFORM for programs that can't build on a
particular platform for technical reasons.
See print/acroread7, which has LICENSE and NO_*_ON_* due to a
restrictive license, but also ONLY_FOR_PLATFORM for i386 since it's an
i386 binary that's available, which is a technical issue, even if it
is because of the license.

See devel/libusb and devel/lwp for packages that have only technical
and not licensing issues.

-- 
        Greg Troxel <gdt@ir.bbn.com>