ccan code breaks older build farm systems
abartlet at samba.org
Thu Jul 7 06:30:58 MDT 2011
On Thu, 2011-07-07 at 13:54 +0200, Michael Adam wrote:
> Hi Rusty,
> my first misconception was that I thought you had added the ccan
> code to samba proper. In that case it is simply our policy to add
> the copyright / license boilerplate commentes to the code files.
> It was not clear to me that the ccan code is considered external
> contributed code whose authoritative copy is not inside the samba
> Anyhow, since you added the ccan code so that it is directly linked
> to virtually all samba3 binaries via the LIB_OBJ collection in
> source3/Makefile.in, to my understanding, you need to relicense
> the LGPL files to GPLv3:
> It says, it is OK to include the LGPLv3 code into the GPlv3+
> project if you relicense the copy of LGPLv3+ code that is
> included in samba under GPLv3.
> Generally, I think we should make it more clear in our code tree which
> code is original samba code and which is external contributed code.
> I think we should not put such external code directly under lib/
> but add a contrib/ folder (or similar) to contain externally
> contributed code.
As to names, a contrib folder is traditionally for code that is hosted,
but not supported by the project. ie, contributed modules that have
external maintainers. Certainly another name could be chosen however.
As you know, we have quite a bit of external code in the tree, which
maintains it's original licence headers and is not in any one location -
we have lib/popt, source4/heimdal, source4/scripting/samba-external and
buildtools/bin/waf*. In most cases we have a README explaining the
original source, but we may be missing some. I would not favour a
wholesale relocation of code to create an artificial line between
'samba-developed' and imported codebases.
I understand how you read the FSF's website suggesting that we have to
change licence headers of imported code. However, this has not been our
practice in the past, may be considered rude by the original authors,
(yes, it happens!) would make it harder to clearly contribute that code
back. As a thought experiment, imagine if we did that to the Heimdal
codebase we also host!
Andrew Bartlett <abartlet at samba.org>
More information about the samba-technical