You probably should back off until I get a chance to merge our stuff in. The "blobs" are gone. The information that needs to be created at compile time to be passed to the library are now simple arrays of build_int_cst_type(integer_type_node,...) integers that become CONSTRUCTORs for arrays.
As for the information coming back... I'll try to look at that, too. With this exception code refactoring out of the way, we should be more visible here again. I figured out you were in central Europe. But you often seem to be at work towards midnight many nights! > -----Original Message----- > From: Jakub Jelinek <ja...@redhat.com> > Sent: Friday, May 2, 2025 13:16 > To: Robert Dubner <rdub...@symas.com> > Cc: James K. Lowden <jklow...@cobolworx.com>; gcc-patches@gcc.gnu.org > Subject: Re: [PATCH] cobol: Fix up exception handling [PR119364] > > On Fri, May 02, 2025 at 12:01:14PM -0500, Robert Dubner wrote: > > All we can do is see what happens. So, please, apply your patches to > > trunk. If you can do that soonish (and since it's not easy to figure > out > > Committed now. > > Note, there is one last thing for the i686 -> x86_64 cross-compiler, > what I've tried to deal with in > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119364#c20 > but that didn't work as other parts of the FE then read the data back > and were expecting to see the host form of the blobs rather than target. > > Dunno if that part is also planned to be changed or if I should just try > harder on that patch and do the reshuffling (host to target) elsewhere. > > > exactly when you sleep, I am hopeful that you can) then I'll build a > patch > > on top of that, and we'll see what we shall see. > > I'm in CEST timezone. > > Jakub
