Hi,
Nuttx is BSD licensed while osmocom-bb's src/target/firmware is GPLv2(or later). Mixing GPlv2 and BSD code is legally possible. As I understand it the whole work becomes GPLv2 when the code is mixed, the BSD part however can be used independently of the GPL part if you strip out the GPL part and the BSD copyright headers have to remain intact.
The problem is that upstream(nuttx) will unlikely accept GPL code as-is for inclusion in nuttx.
However there is a misc directory for software made under different licenses such as applications or drivers(there is one GPL driver for the rtl8187x wifi chipsets). To use the driver the user is expected to run a script that install the driver in nuttx normal source code directory.
The list of files touched by the patches mades on top of nuttx and their licenses is listed here: http://bb.osmocom.org/trac/wiki/nuttx-bb/code-audit
Basically the GPL parts are composed by : * the irq * the timer * the clock * the uart
some of the related files have the following authors(can be combined together): * Harald Welte * Stefan Richter * Ingo Albrecht
I guess Stefan Richter and Ingo Albrecht will be hard to reach.
So I wonder what's the best thing to do.
Denis.
Hi Denis,
On Tue, Feb 28, 2012 at 10:17:37PM +0100, Denis 'GNUtoo' Carikli wrote:
some of the related files have the following authors(can be combined together):
How did you do the analysis? Just by looking at the header of the file, or actually analyzing the author name information in the git commit logs of those files? The latter is probably more precise. Of course you can skip any trivial one-line changes as they are not copyrightable.
- Harald Welte
As indicated before, I don't mind re-licensing under BSD-style license for the core drivers. The actual GSM protocol stack is not to be relicensed, but this is not what we're talking about anyway.
- Stefan Richter
I havent' heard of im in some time, sorry.
- Ingo Albrecht
He's a friend here in Berlin, and I believe he has already verbally agreed to such relicensing.
How did you do the analysis? Just by looking at the header of the file, or actually analyzing the author name information in the git commit logs of those files? The latter is probably more precise. Of course you can skip any trivial one-line changes as they are not copyrightable.
I just looked at the header, that was quicker.
Denis.
Hi Denis,
On Thu, Mar 01, 2012 at 09:15:35PM +0100, Denis 'GNUtoo' Carikli wrote:
How did you do the analysis? Just by looking at the header of the file, or actually analyzing the author name information in the git commit logs of those files? The latter is probably more precise. Of course you can skip any trivial one-line changes as they are not copyrightable.
I just looked at the header, that was quicker.
yes, but also much less precise. doing a git log on a file and grep'ing for the author like:
git log --follow layer1/prim_fbsb.c | grep ^Auth | sort | uniq
is not that hard. Of course it only works in a repository that has the complete history.
Regards, Harald
Harald Welte wrote:
I just looked at the header, that was quicker.
yes, but also much less precise. doing a git log on a file and grep'ing for the author like:
git log --follow layer1/prim_fbsb.c | grep ^Auth | sort | uniq
is not that hard. Of course it only works in a repository that has the complete history.
The data set can be reduced by looking only at the latest state of the code:
git blame file.c | \ sed 's,[^ ]+ (([^)]*) [0-9]{4}-[0-9]{2}-[0-9]{2} .*,\1,' | sort -u
..since this is what would be contributed to upstream anyway.
//Peter
baseband-devel@lists.osmocom.org
-
Denis 'GNUtoo' Carikli -
Harald Welte -
Peter Stuge