* Ralf Corsépius: > Am 06.06.22 um 16:38 schrieb Petr Pisar: > >> I believe that glibc payed a special attention to provide both >> ABIs. But that does not apply to other libraries. Maybe it's not so >> important problem. Proprietary software tends to bundle the >> libraries, externally relying only on kernel and glibc. (X libraries, Motif, glib are usually not bundled, either, and there are a couple mor examples. It's likely that there is at least some time_t or struct timeval/timespec exposre in all of these otherwise ABI-stable libraries.) > How about dlopened libraries? I'd assume these can become problematic. dlsym calls need to be taught to use the time64 aliases. Symbol versioning isn't used. Maybe we could add a time64 version of dlsym and dlvsym that knows about the aliased symbols, but we really shouldn't be doing new development work for 32-bit architectures. It doesn't benefit Fedora, and most embedded 32-bit users do not use glibc, either. However, there was a strong push from a few interested parties who funded the initial development (and at least one party is still involved and addresses any fallout from the changes). I'm not really happy how things turned out, but at least there is no immediate harm to Fedora's i686 compatibility packages (beyond the occasional bug, but as I mentioned, someone is still around to fix them). Thanks, Florian _______________________________________________ devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure
