https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87003
--- Comment #2 from Tom Tromey <tromey at gcc dot gnu.org> --- I don't really know the best thing to do. I see your point about graceful failure being a useful feature, in cases where the result of some gcc-jit function is passed as an argument to another one. Maybe there are some arguments that truly cannot be NULL? If any exist I think it would be fine to mark them as nonnull and have the implementation simply assert. Perhaps there aren't many of these though. Another thing that would be useful would be a section in the manual explaining the null-handling approach. I see there is some text about this but perhaps it could be made more prominent, like having a whole section just dedicated to this topic?