On Mon, Dec 06 2021, Daniel Dickman <didick...@gmail.com> wrote: >> On Dec 6, 2021, at 5:44 PM, Jeremie Courreges-Anglas <j...@wxcvbn.org> wrote: >> >> On Sun, Dec 05 2021, Christian Weisgerber <na...@mips.inka.de> wrote: >>> I ran another amd64 bulk build with base clang updated to LLVM 13 >>> and --allow-shlib-undefined as the default again (~patrick/llvm.v6). >>> I also put in a tentative fix for security/nss. >>> >>> Failure logs: >>> http://build-failures.rhaalovely.net/amd64-clang/2021-12-04/ >>> >>> Triage: >> >>> emulators/nono -Werror,-Wunused-but-set-variable >> >> Looks as simple as dropping -Werror. ok? > > > Why not make it a bit more precise and just add something > like -Wno-unused-but-set-variable ? > > I guess it’s possible other types of warnings could be hit though…
Yep. Other types of warnings could be hit after the one fixed by -Wunused-but-set-variable: in the same run, at the next nono update, the update of a dependency, the next base-clang update, etc. -Werror doesn't scale well given the many platforms, compilers and environments that can be found in OpenBSD land and elsewhere. I would find it helpful if people removed -Werror whenever they see it and it doesn't result in a mess of patches. > Or better yet figure out if the code can be fixed or if the compiler is > spitting out a false positive? While both *can* be interesting I suspect it's not the case here. Note that you or Gonzalo or anyone else are free to fix this properly upstream (or in the ports tree if there's an actual problem). > Either way ok on the original update to 0.2.4. I'm not sure what you mean, I don't have an update to 0.2.4 at hand. Do you mean that the patch still fixes the build with clang 13 even after updating to nono-0.2.4? -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
