[Buildroot] Analysis results for 2018-10-09

Matthew Weber matthew.weber at rockwellcollins.com
Fri Oct 12 03:57:31 UTC 2018


Thomas,


On Thu, Oct 11, 2018 at 1:50 AM Thomas Petazzoni
<thomas.petazzoni at bootlin.com> wrote:
>
> Hello,
>
> On Wed, 10 Oct 2018 16:25:41 -0500, Matthew Weber wrote:
>
> > Adam, did you have something in the works for this one?  I seem to
> > remember some discussion on the one "whole" thread above related to
> > breaking the dependency.  I'd make the change but I don't have a valid
> > test case.
>
> Well, Arnout statement was that the build dependency is not necessary
> because python3 can dlopen() at runtime the libuuid.so library. Indeed
> the code in Python3 is here to do that, but as I explained in
> http://lists.busybox.net/pipermail/buildroot/2018-September/231060.html,
> it doesn't seem to work, at least in the Buildroot context.
>

I dug through this a bit more and found that the c library is
providing the implementation for uuid.uuid1() when uuid_generate_time*
is probed for in libraries using ctypes.  I believe that the
util-linux libuuid would come into play if the extension module
approach worked (all the configure tests currently fail which disable
use of python3-3.7.0/Modules/_uuidmodule.c).

I believe the right fix for now is cleaning up the current approach
(removing the ability to enable/disable uuid and the util-linux
dependency) as it looks like python can support uuid by default with
the current glibc.  My guess is we'll see build failures with the
other std libraries....

Matt



More information about the buildroot mailing list