Issue w/ compiling on RHEL 7

Question

Issue w/ compiling on RHEL 7

eltatio opened this issue 5 years ago · 3 comments

I have issues w/ compilation on RHEL 7 and anaconda/python 3

Below some of the errors preventing proper compilation :

/local/laurent/anaconda3/include/python3.7m/pybind11/numpy.h:317:5: erreur: ‘is_trivially_copyable’ is not a member of ‘std’
std::is_trivially_copyable,
^
/local/laurent/anaconda3/include/python3.7m/pybind11/numpy.h:317:5: erreur: ‘is_trivially_copyable’ is not a member of ‘std’
/local/laurent/anaconda3/include/python3.7m/pybind11/numpy.h:317:33: erreur: patron de l'argument 2 est invalide
std::is_trivially_copyable,
^
/local/laurent/anaconda3/include/python3.7m/pybind11/numpy.h:949:49: erreur: ‘is_pod_struct’ is not a member of ‘pybind11::detail’
struct format_descriptor<T, detail::enable_if_t<detail::is_pod_struct::value>> {
^
/local/laurent/anaconda3/include/python3.7m/pybind11/numpy.h:949:49: erreur: ‘is_pod_struct’ is not a member of ‘pybind11::detail’
/local/laurent/anaconda3/include/python3.7m/pybind11/numpy.h:949:72: erreur: patron de l'argument 1 est invalide
struct format_descriptor<T, detail::enable_if_t<detail::is_pod_struct::value>> {

gcc version :
g++ (GCC) 4.8.5 20150623 (Red Hat 4.8.5-28)

Thanks for hint to correct.

Laurent.

Answer 1 · 2019-10-24T12:15:41.000Z

Hmm obviously an issue with pybind11. For a quick fix try gcc5 and I will check it myself more thoroughly. It is definitely a compiler version issue but the build should take that into account. What was your installation command?

Answer 2 · 2019-10-24T16:51:23.000Z

Hello Codie, Thanks for support Finally it worked with g++-5 👍 In the mean Time we tested also lasso with simple models. Did you also consider the possibility to write d3plot files? If not, would it be possible to develop this as a service ? Thanks by avance Laurent Guérin... Le jeu. 24 oct. 2019 à 14:15, Codie (蒂诺) <notifications@github.com> a écrit :

…

Hmm obviously an issue with pybind11. For a quick fix gcc5 should work and I will check it myself more thoroughly. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#62?email_source=notifications&email_token=AK7ENLRCBJFGWLQ5FORXXSTQQGGW5A5CNFSM4JEAYFZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECEZ47Q#issuecomment-545889918>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AK7ENLXPTVQAGNE4O6SHLBDQQGGW5ANCNFSM4JEAYFZA> .

Answer 3 · 2019-10-25T08:23:14.000Z

Hey Laurent,
it indeed would be possible for me to implement a d3plot writer. We didn't need it yet but if you are seriously interested you can contact me under codie@lasso.de.

I will leave the issue open since it ought to compile under gcc 4.8. My guess is the installation process somehow used a newer pybind11 version which now requires at least gcc5 or to be more precise the c++14 standard.