Subject: Re: [boost] [builds] broken
From: Tom Kent (lists_at_[hidden])
Date: 2017-03-05 12:49:05
On Thu, Mar 2, 2017 at 9:59 PM, Rene Rivera via Boost <boost_at_[hidden]
> wrote:
> On Thu, Mar 2, 2017 at 9:52 AM, Raffi Enficiaud via Boost <
> boost_at_[hidden]> wrote:
>
> > Le 02/03/2017 à 16:25, Rene Rivera via Boost a écrit :
> >
> >> On Thu, Mar 2, 2017 at 9:22 AM, Raffi Enficiaud via Boost <
> >> boost_at_[hidden]> wrote:
> >>
> >> Le 02/03/2017 à 16:20, Rene Rivera via Boost a écrit :
> >>>
> >>> On Wed, Mar 1, 2017 at 5:23 PM, Raffi Enficiaud via Boost <
> >>>> boost_at_[hidden]> wrote:
> >>>>
> >>>> Hi all,
> >>>>
> >>>>>
> >>>>> On develop, I have the following error for gcc/linux and clang/osx:
> >>>>>
> >>>>> 01-Mar-2017 23:58:05 libs/python/build/Jamfile:155: in
> >>>>> lib_boost_numpy
> >>>>> 01-Mar-2017 23:58:05 *** argument error
> >>>>> 01-Mar-2017 23:58:05 * rule lib ( names + : sources * :
> >>>>> requirements *
> >>>>> : default-build * : usage-requirements * )
> >>>>> 01-Mar-2017 23:58:05 * called with: ( : numpy/dtype.cpp
> >>>>> numpy/matrix.cpp
> >>>>>
> >>>>>
> >>>>
> >>>> How are you configuring python in your user-config.jam?
> >>>>
> >>>>
> >>> I am not touching anything about user-config.jam: the failure happens
> on
> >>> a
> >>> build machine (specific build user, supposedly clean). If there is a
> >>> user-config.jam, then I do not know how it landed there :)
> >>>
> >>
> >>
> >> OK.. Which means that the version number for whatever python it picks up
> >> is
> >> undetermined and blank. I'll have to figure out why that is.
> >>
> >
> > FYI: the Linux/OSX machines have the apt/OS python installed, and
> > python3.4.
>
>
> OK.. It turns out to have been a simple forgetting of some parens. Note
> that bootstrap generates a project-config.jam file. So you do have one. You
> just don't see what it is :-) But it does print out in the CI logs what it
> detects as far as Python is concerned. Anyway.. It's fixed. And will be
> reflected whenever stuff updates to latest develop.
I think this is related, but not sure...all the regression runs I start
(Linux + Windows) are currently failing with:
boost-test(COMPILE) "python/select_holder" :
"libs/python/test/select_holder.cpp"
boost-test(RUN) "python/select_from_python_test" :
"libs/python/test/select_from_python_test.cpp" "libs/python/src/conve
rter/type_id.cpp"
boost-test(COMPILE) "python/select_arg_to_python_test" :
"libs/python/test/select_arg_to_python_test.cpp"
boost-test(COMPILE_FAIL) "python/raw_pyobject_fail1" :
"libs/python/test/raw_pyobject_fail1.cpp"
boost-test(COMPILE_FAIL) "python/raw_pyobject_fail2" :
"libs/python/test/raw_pyobject_fail2.cpp"
boost-test(COMPILE_FAIL) "python/as_to_python_function" :
"libs/python/test/as_to_python_function.cpp"
boost-test(COMPILE_FAIL) "python/object_fail1" :
"libs/python/test/object_fail1.cpp"
error: at /var/boost/run/boost_bb/src/kernel/modules.jam:107
error: Unable to find file or target named
error: '/boost/python//boost_numpy'
error: referred to from project at
error: '../libs/python/test'
at the end of the results/bjam.log file. This was causing develop to fail
for awhile, and now appears to have migrated to master as well.
Tom