[RFC] X.Org minimum requirements for Autotools policy review

Daniel Stone daniel at fooishbar.org
Fri Oct 25 00:24:12 CEST 2013


On 24 October 2013 22:17, Alan Coopersmith <alan.coopersmith at oracle.com> wrote:
> On 10/24/13 11:17 AM, Gaetan Nadon wrote:
>> GPLv3 Autoconf licensing still an issue?
>>     A couple of years ago, some platforms declared they were unable to
>> ship any
>>     software package if it was licensed under GPLv3. This is the case for
>>     Autoconf 2.62 and I recall XCB had to rollback to 2.60. Anyone knows
>> if this
>>     has been rosolved?
> This is not a problem for my employer, but I think it remains a problem for
> certain vendors who sell consumer products to which they don't want to give
> purchasers the ability to recompile & reinstall the software on the devices,
> and I don't think there is any resolution possible there, as it's simply
> incompatible with their business choices.

Right.  Even though autoconf is very clear about the GPL not leaking
from build system infrastructure into the code it actually builds (or
even the finished product: I believe the output is explicitly freely
redistributable with no catch, rather than GPLed itself), it still
won't quite fly.  A lot of people have taken the option to completely
exclude all GPLv3 products from their chain entirely, so as to have no
doubt whatsoever.

As newer versions of autoconf et al with GPLv3 become more embedded
(ha), I think that's going to change.  Anecdotally, I'm seeing a lot
less 'nothing GPLv3 anywhere ever' around already.

I was opposed originally, but if it actually enables us to do more
useful things with the build system, then go for it.  I do remember
some very useful changes in automake 1.11, so even just being able to
have that is probably enough to make it worth bumping the autoconf
minimum version.


More information about the xorg-devel mailing list