https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119889
--- Comment #8 from Jürgen Reuter <juergen.reuter at desy dot de> --- > > Hi Harald, > as I reported this to NAG and Intel as well, I got the following reply from > Malcolm Cohen from NAG: > "I note that there is an error in your example at line 56: it has > procedure(dlsym), pointer :: tmp_fun > however, it is not being pointer-assigned dlsym, but the result of > referencing dlsym. Therefore it should be something like > procedure(), pointer :: tmp_fun" > > After correcting this, both Nagfor and Ifx compile, as in gfortran. > Cheers, > Jürgen One more thing that Malcolm wrote: "After correcting this mistake, I note that if you give the function result an explicit interface, the panic will not happen. (It may also a good idea to give "tmp_fun" the same explicit interface as below, but that is not necessary.) For example, instead of procedure(), pointer :: f_funp at line 28, the declaration interface subroutine subroutine end subroutine end interface procedure(subroutine), pointer :: f_funp will work. "
