On Thu, Aug 6, 2015 at 9:01 AM, Mark Asselstine <[email protected]> wrote: > On August 5, 2015 21:11:35 Bruce Ashfield wrote: >> On Wed, Aug 5, 2015 at 4:29 PM, Mark Asselstine >> >> <[email protected]> wrote: >> > Ran in to a build error: >> > distutils.errors.DistutilsError: Could not find suitable distribution for >> > Requirement.parse('pbr') ERROR: python setup.py build execution failed. >> > >> > Fix by DEPENDing on python-pbr. >> >> This used to be on purpose. If you added python-pbr to DEPENDS the images >> would fail to assemble due to QA errors and overlapping files. >> >> If you say that you've successfully assembled and booted the images and pbr >> isn't acting up, I'll merge this .. but otherwise, I'm eyeing the >> change for more >> testing. > > Some quick build testing this morning and I can confirm that both this package > and the python-oslo.serialization packages *will not* build without pbr. For > this package we can see that upstream has converted to building with pbr so > the build failure in the absence of pbr is easily explained: > > --- > commit 915b2b604c0e9b97e813711c035dfd13682fb6df > Author: Robert Collins <[email protected]> > Date: Tue May 5 10:45:22 2015 +1200 > > Switch to pbr as a build system > ... > --- > > Similarly for python-oslo.serialization we can see commits in the upstream > repo which point to the required pbr dependency: > > --- > commit a7bade117a8f8297955c0cb8adb6757d30f02e1b > Author: Doug Hellmann <[email protected]> > Date: Fri Oct 24 09:29:18 2014 -0400 > > Add pbr to installation requirements > > Add pbr to the list of installation requirements so that it is installed > via pip before this library is installed, instead of with easy_install. > This avoids issues like Bug #1384919, and ensures that projects that use > this library as a dependency are properly installed. > > Change-Id: I6c155370dbd01fe4748d5137bdf288e8d3e1a67e > --- > I can't easily identify the change that made pbr mandatory but the build > failure is 100% reproducible so there is definitely a change which prevents us > from treating this as optional. > > I can also confirm that with these changes there exists no QA errors that > prevent the images from being constructed. I did builds from scratch and > others out of sstate without issue. I would be more than happy if Li would > like to confirm this by doing a build with this series applied.
Good enough for me. I've staged the series here and will push it shortly. It can't make things worse .. so in it goes! Bruce > > Mark > >> >> Bruce >> >> > Signed-off-by: Mark Asselstine <[email protected]> >> > --- >> > >> > meta-openstack/recipes-devtools/python/python-testtools_1.8.0.bb | 4 ++++ >> > 1 file changed, 4 insertions(+) >> > >> > diff --git >> > a/meta-openstack/recipes-devtools/python/python-testtools_1.8.0.bb >> > b/meta-openstack/recipes-devtools/python/python-testtools_1.8.0.bb index >> > 80a7489..a58c91a 100644 >> > --- a/meta-openstack/recipes-devtools/python/python-testtools_1.8.0.bb >> > +++ b/meta-openstack/recipes-devtools/python/python-testtools_1.8.0.bb >> > @@ -17,4 +17,8 @@ S = "${WORKDIR}/${SRCNAME}-${PV}" >> > >> > inherit setuptools >> > >> > +DEPENDS += " \ >> > + python-pbr \ >> > +" >> > + >> > >> > RDEPENDS_${PN} += "python-extras" >> > >> > -- >> > 2.1.4 >> > >> > -- >> > _______________________________________________ >> > meta-virtualization mailing list >> > [email protected] >> > https://lists.yoctoproject.org/listinfo/meta-virtualization >> >> -- >> "Thou shalt not follow the NULL pointer, for chaos and madness await >> thee at its end" > -- "Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end" -- _______________________________________________ meta-virtualization mailing list [email protected] https://lists.yoctoproject.org/listinfo/meta-virtualization
