2006 is almost gone, more pure death in 2007?

[Floid]

dammy wrote:

This will allow greater freedom on doing ports of other open source software that would not be possible (ie license conflicts) with the current x86/PPC AROS kernel.

How does the kernel license influence the licensing of applications?

How does LGPLing the kernel make it easier to reuse any code from that one major open-source kernel project which comes bound by the terms of the more restrictive GPL?

If the kernel is modular, with clearly defined interfaces, and drivers are distributed separately from it, does the license really matter at all?  (Even if individual users  'taint' their kernels, as long as they don't redistribute that combination as a single package they're pretty well out of the grey area.  Same thing goes for ugly proprietary Linux drivers, like the nVidia blobs, though the concern there is that RedHat et al would like to be able to legally ship single packaged products that actually work on anyone's hardware.)

I'm not going to complain about a less obfuscated, less restrictive license --- the LGPL specifically avoids the OMG-you-can't-distribute-this-it's-tainted problem, and tainting doesn't seem to have ever stopped any jack*** manufacturers from offering blob-only support --- but I'm not sure this solves what you think it's solving.

The BSDs have some pretty good driver code, anyway.

[Floid]

pixie wrote:

Can you tell in plain English what LGPL license is better then the present one? Does it attract more developers?

The Lesser GPL will definitely allow for greater license-compatibility of code from going outwards (since there are many LGPL projects in the world, and provision 3 of the LGPL allows LGPL'd code to be GPL'd -- meaning further restricted, and thus compatible with inclusion in the Linux kernel tree, say).

But... since we're only talking about the AROS kernel being LGPL'd, that doesn't necessarily mean it contains anything interesting for interoperability in the real world, like filesystems or whatever.

Inbound, well, some big interesting driver stacks (ALSA, ferinstance) are available under the LGPL, so including (derivatives of) them would be less of a brainteaser... except for the fact that the kernel isn't monolithic, so depending how you define linking, it may or may not even matter.


The AROS license isn't a particularly bad license -- the provisions that require a paper (electron?) trail in the license notices aren't a bad idea, and same had to be adopted ad-hoc by Linux to allow graphing of who claimed authorship/rightsholdership over submitted code...  But being different does block you from incorporating anything intended for the GNU/"copyleft" community, which by definition exists to ensure the library of copylefted stuff grows.


Since one thing AInc. did manage to do before going twirling, twirling, twirling towards victory was respectfully ask that AROS code be as fricking inconvenient for normal humans to review as possible, it's hard to say how much of anything in the tree has already been dual-licensed AROS and GNU by its authors anyway.
 (See here; of course, this seemed reasonable at the time, but we didn't know the 'community' would be at an impasse for the next five years, and that AROS core would not, say, be coughing up source snapshots as part of their release cycle.)

We're all so **** good at community-building, it's a wonder anyone's found time to ship product!