[PATCH] new applet: nmeter

Rob Landley rob at landley.net
Wed Aug 23 20:41:21 UTC 2006


On Tuesday 22 August 2006 9:49 pm, Rich Felker wrote:
> On Tue, Aug 22, 2006 at 05:17:39PM -0400, Rob Landley wrote:
> > If you want to hold off on merging this patch after GPLv3 ships, we'll 
> > probably have gone GPLv2 only at that point.  But right now, we're holding 
> 
> May I ask one final time, WHY? What advantage do you get by preventing
> people from using the code under GPLv3 as well as GPLv2?

1) Right now we can't merge GPLv2 only code (which is a real world problem for 
us since it means that not only can we not take stuff like the diethotplug 
code, but the portion of the kernel headers we sucked into loop.c are legally 
questionable.).  We also can't merge GPLv3 only code, and never will be able 
to, not that this has ever mattered.  So having both licenses limits the 
number of sources we can take from.

2) Since the Linux kernel rejected it, it seems unlikely to have particularly 
widespread adoption, so the upside of going through _any_ pain to keep 
compatability with it is nonexistent.  It's become just another incompatible 
license ala Sun's CDDL or whatever that is (as if we had a shortage of them).  
For BusyBox to be dual licensed under CDDL would be about as relevant.  The 
one we _care_ about is GPLv2.

3) GPLv3 attempts to place limits on usage, not just distribution.  This makes 
it a more restrictive license than GPLv2, leads to a forest of gotchas (like 
the the need to distribute the private key you sign an RPM with in order to 
comply with the license terms) only some of which are likely to be found 
before the final version goes out.  It also makes it particularly distasteful 
for the embedded market, although since we're never dropping GPLv2 (and 
that's the only one we're ever going to enforce) it's also largely 
irrelevant.

4) The GPLv2 ain't broken, and we're never dropping GPLv2.  Why do we _need_ 
two licenses?  (I realize that people keep saying "but what if v2 breaks", 
like it's a dish on a shelf somewhere.  What if the license Windows is under 
breaks?  Think that's likely?  The GPLv2 has been out for 15 years now.)  
Sheer simplification says dumping the one we don't use and have no plans to 
enforce is the obvious move.

> Are you 
> worried that someone will make a derived work and change the "v2 or
> later" to "v3 only" and that you'll want to incorporate their changes
> into BB but be unable to do so?

I fail to see how it's a plus, but I'm not particuarly worried about it.  I 
don't expect there to _be_ much v3 code from anywhere except the FSF, and 
most FSF code is not a good match for BusyBox.  And how this would differ 
from any other external code that's v3 only (or any other license terms that 
we just can't use) is anybody's guess.

> Unless this is your concern, there is 
> absolutely no reason to remove the "or later" except to make people's
> lives difficult for the sake of political bickering.

You have a talent for being wrong a lot.

Right now we can't use the diethotplug Eric ported to busybox.  That's real, 
and GPLv3 currently isn't.

Right now it's quite possible that a random kernel developer could come to us 
and say "yank loop.c, you can't distribute that under GPLv3".  (Since the 
header defines an API that one's probably ok under "scenes a faire", but 
there could be other instances of this sort of thing in the code because 
we've always assumed that the Linux license and the busybox license were the 
same thing.)

> The last thing we 
> need is for this sort of pettiness to fragment GPL'd software by
> splitting it into two mutually incompatible "v2 only" and "v3 only"
> corpuses of code that can't be used together.

I think the Linux kernel already managed that, and in case you hadn't noticed 
I've considered what the Linux kernel developers do to be a good model for 
about eight years now.  I'm sure I've mentioned it more than once here. :)

> As far as I can tell, v3 has its place but it's not right for BB.

HOW IS THIS A DEFENSE OF KEEPING A DUAL V3 LICENSE ON BUSYBOX?????

> However it's ridiculous to prevent people who need v3 from
> incorporating BB code into their projects.

A) By that logic we should be in the public domain.

B) There are currently zero people on the entire planet who need v3, because 
it doesn't exist yet so no projects can be licensed exclusively under it.

C) One advantage of dropping v3 _before_ it ships (or really soon afterwards) 
is not giving anybody trying to decide whether or not v3 is a good license 
for their project the mistaken impression that future versions of BusyBox 
will provide code they can use under v3 forevermore.  This is, most likely, 
not going to be the case.

External projects are always welcome to make bad decisions.  This is not my 
problem.  GPLv3 is a brand new license that has nothing to do with v2, and we 
are not actually required to issue new code under that license.

That said, I'm still leaning towards waiting until GPLv3 actually ships before 
making a final decision on this, and I'm definitely keeping Eric Andersen in 
the loop.  (Of course waiting for GPLv3 to ship assumes its schedule isn't 
like Duke Nukem Forever or Windows Vista...)

> Rich

Rob
-- 
Never bet against the cheap plastic solution.



More information about the busybox mailing list