Created attachment 1414831 [details] preprocessed Description of problem: compiled with g++ -o logsumexp.cpython-36m-x86_64-linux-gnu.os -c -g -std=c++1z -O3 -march=native -fvisibility=hidden -flto -DNDEBUG -ffast-math -fPIC -I/home/nbecker/.local/include/xtensor-blas/flens -I/home/nbecker/.local/include -I/usr/include/eigen3 -I/home/nbecker/.local/lib/python3.6/site-packages/numpy/core/include -I/usr/include/python3.6m logsumexp.cc Version-Release number of selected component (if applicable): How reproducible: 100% Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Thanks; I'm able to reproduce with both trunk and gcc-7-branch, with: ./xg++ -B. -c /tmp/foo.cc -std=c++1z I'm working on simplifying the reproducer.
I ran into two (apparently different) internal compiler errors on the attachment. Issue (a): unexpected expression ‘I’ of kind template_parm_index appears to be a duplicate of https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84785 (I saw it on gcc-7-branch with r258233, and I no longer see it after updating to r259103, so presumably fixed by r258409 on that branch). Issue (b) is the failure of this assertion in gcc/cp/call.c: 8243 gcc_assert (!force_elide); which still seeing on trunk. I've filed this one upstream as: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85215
PR c++/85215 was fixed 2018-04-05 by r259123 PR c++/84785 was fixed 2018-03-10 on trunk by r258407 on gcc-7-branch by r258409 (will be in 7.4.0)
Both PRs fixed in current F28 gcc.