https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98301
--- Comment #7 from Steve Kargl <sgk at troutmask dot apl.washington.edu> --- On Mon, Feb 22, 2021 at 06:57:03PM +0000, sgk at troutmask dot apl.washington.edu wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98301 > > --- Comment #6 from Steve Kargl <sgk at troutmask dot apl.washington.edu> --- > On Mon, Feb 22, 2021 at 04:36:50AM +0000, jvdelisle at gcc dot gnu.org wrote: > > > > --- Comment #5 from Jerry DeLisle <jvdelisle at gcc dot gnu.org> --- > > Steve, if you think this does it. I will get it ready to commit for you. > > Does > > it also need to be back ported? > > > > Well, I think it is correct for at least what I've been > able to test! I don't use co-arrays, so do not have > OpenCoarray or the shared-memory WIP installed. For, > -fcoarray=lib or -fcoarray=shared, function calls are > generated. These are currently unimplimented and need > to be implemented in such a manner as the function calls > are aware of the underlying coarray implemention. I > don't remember the details, but I recall that the images > need to communicate about the seeds. > Oh, one last item. This patch will likely break gfortran+opencoarray because _gfortran_caf_random_init is missing, and therefore will prevent linking of the program.
