[BusyBox] Compiling busybox with newlib [PATCH]

[Rob Landley]

On Saturday 05 March 2005 09:16 pm, Shaun Jackman wrote:
> On Wed, 2 Mar 2005 12:43:21 +1100, Glenn L McGrath <bug1 at iinet.net.au> 
wrote:
> > hmmm... its a fine line your walking, to the best of my knowledge...
> >
> > If you have not released your GPL derived work, your work is still under
> > the GPL, its just that the GPL doesnt demand anything much of you yet.
>
> It's a fine line, granted, but that line exists for a good reason.

I wouldn't say you're walking a fine line.  You're welcome to do any in-house 
research you want, and as long as you never distribute the resulting binaries 
you're not breaking any license agreements.

However, if you ever intend to distribute the result at some point, you can 
ONLY distribute the result under the terms of the GPL.  So you're either 
creating software that will be licensed under the GPL, or you're engaged in a 
pure research project that can never and will never be distributed under any 
terms at all.  These are your only two options.

> There's a non-trivial amount of work involved in bringing all the
> copyright holders and licenses of a large project into alignment such
> that the work as a whole can be redistributed.

This is not our problem. :)

> At this point, I'm more 
> interested in completing the technical work than the legal work.

With the result that you're almost certainly going to "ship to a shelf", as my 
old co-workers used to say back at IBM...

> > You only ever have to supply the source code to those you have
> > provided binaries to, so you arent obligated to share your source with
> > the community that created the original work. But personally i think its
> > better to embrace the community rather than hide from it.
>
> I recently reached the point where I thought the work I had completed
> was at a point where it was useful to others and should be given back
> to the community I started submitting patches to BusyBox only a few
> days ago, which stemmed this conversation in fact. I am not trying to
> hide by from the community, but I am bound by the same legal system
> that enforces the GPL. I cannot give away work to which I don't have
> the rights.

Understood.  We're not criticizing.  Despite my criticism of your patches I do 
see valuable work here.  I'm just picky about what goes into busybox.  
(Erik's the final judge you have to worry about, though.)

> > I hope you can convince your employer that the long term benefits of
> > co-operative development far outweigh the short term costs.
>
> If it were only my employer, it would be a simpler case. I also have
> to consider the rights of the copyright holders of libraries licensed
> for the project. In the end I will make all the licenses come together
> and "play nice", but in the mean time I will submit patches to the
> individual open source projects from whose excellent work I have
> drawn.

It has long been legal to run GPL software on a proprietary OS, because the OS 
is not a derivative work of the GPL software.  If newlib and libgloss, 
between the two of them, constitute an embedded OS platform, then an 
_unmodified_ version of busybox running on top of them probably won't get you 
into trouble.  (Although there's definitely issues with bundling them 
together into the same binary, since that binary IS a derivative work of 
busybox.  It's hard to find a good IP lawyer who's up to speed on open source 
issues, but OSI knows a few.)

When you have to modify busybox to get it to work on newlib and libgloss, that 
opens a whole new can of worms as well.  We're understandably reluctant to 
accept patches that do nothing but modify our GPL code to work with non-GPL 
code so there are fewer questions about the other code's derived work status.  
If there's no derived work, why does our code need to be modified?

> Regards,
> Shaun

Rob