https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119364
--- Comment #19 from Jakub Jelinek <jakub at gcc dot gnu.org> --- (In reply to Robert Dubner from comment #18) > The revised code will be based on passing pointers to lists of integers. > > (I have managed to talk Jim out of his desire to pass structures created > with GENERIC to the library code. He thinks I am a sissy for not wanting to > do it, but I have gotten mulish about it.) > > Those lists of integers are being established as static arrays of type > integer_type_node with constructors using > build_int_cst_type(integer_type_node,) to set each element. That will be nice, but the problem is that we want to release GCC 15.1 this month, ideally branch it of from master next week. So, the intent was to get COBOL working for the release reasonably well and further development can be done on the trunk (to be GCC 16) and perhaps some safe parts backported for 15.2 or 15.3 (though ideally nothing changing ABI). So, from my POV (and with GCC release manager hat on), it would be great if e.g. the libgcobol libquadmath patches could be reviewed and committed, maybe (lower priority) this too, so that people can at least try some cross-compilers.
