On Wed, Oct 15, 2008 at 11:20:06AM -0400, Frank B. wrote:
Suppose, for example, I have some code that’s more Cognitive-Radio-related
than Software-Defined-Radio, really. It uses Orange <
http://www.ailab.si/orange>, which is GPL but not assigned to FSF. At least
for now it can’t go into the GNU Radio tree. Probably it never will.
Do you mean that your code can’t go in the repo, or that Orange can’t
go in the repo? I see no problem having code in the repo use Orange.
With regard to the code that’s headed for the trunk, we need to
consider the impact of introducing new external dependencies – an
issue of keeping customers happy, not one of licensing. Code on the
trunk needs to be properly autoconf’d of course. The concern about
new external dependencies doesn’t apply to dev branches.
It’s also true that this (my) code is experimental and provisional. It’s
nothing more than a steppingstone. (The obvious place to put it would be a
developer branch, but that’s part of the tree.) Perhaps I doubt whether, in
its current form, it should go into the tree. But that’s no reason not to
make it visible and easily accessible, if only as a scaffolding for later
code destined for the tree.
That would be an excellent use of a developer branch.
In general, no one should be expecting code in a dev branch to be
fully sorted out, or for that matter, even compile.
There is also some quantity of code which is useful and usable right now,
but which doesn’t currently fit well with the unified installation procedure
for the GR tree. I’m thinking here primarily of Linux audio subsystem code
that uses scons.
No problem with having this on a dev branch.
It seems obvious there has to be a place for GNU Radio code that’s GPL but
will not be assigned to FSF, with certainty. I believe there should also be
a place for code whose status is uncertain – in short, a place with
minimal obstacles to publishing early and often.
I believe that’s one of the CGRAN goals.
Eric
–
All growth occurs at the border of order and chaos…