On Wed, 19 Feb 2025 12:55:03 +0100 Matthias Klose <d...@debian.org> wrote:
> libgcobol/ChangeLog > * Makefile.in: New file. > * acinclude.m4: New file. > * aclocal.m4: New file. > * configure.ac: New file. > * configure.tgt: New file. > > I had updated the configure.tgt, please find it attached here again. It seems we missed your updated patch and invented our own solution, such as it is. Sorry for the misunderstanding. I certainly never meant to ignore you. We know we're currently being too restrictive by limiting host and target to x86_64 and aarch64. But your patch as supplied is also insufficient, so we attempted to generalize it. I would like to do this in a way you approve of. Let me lay out the requirements as of today. For both host and target: 1. support for _Float128 2. support for __int128 My read is that the 2nd requirement is the more restrictive in terms of supported architectures. 6.12 Additional Floating Types indicates that software support is often provided for machines that lack 128-bit floating-point instructions. IIUC --enable-multilib produces separate directories for the built compiler and libraries for 64 and 32 bit on some architectures, x86_64 for sure, but it seems not for aarch64. Color me confused on that one. Because we can't build a 32-bit compiler or library (failing require #2), in the presence of -m32 we disable it in libgcobol/Makefile.in with a $(info) message. I don't think that's an approved tactic; I think the test is supposed to happen in configure.tgt. But as of now that's not possible because the information isn't passed to that script. We know only the target, not the options. > also disabling cobol in the toplevel configury seems to be strange. I think you're referring to disabling the 'cobol' language in the root configure.ac, rather than gcc/configure.ac? We're not sure what the right way is. Happy to oblige. I have an oblique comment in this vicinity. With the goal of testing (the disabling of) 32-bit support during configuration, I was brought up short and up to date at the same time: Linux distributions and cloud vendors have dropped 32-bit offerings. How time does fly! It would seem the intersection of C++14 and 32-bit is narrow indeed. Possible, perhaps, but quickly vanishing except from museums. The cfarm has many 64-bit offerings that are alien to us, and doubtless some lacking the above requirements. I will apply for an account there. --jkl
