From d90232565e55e4bbdf0cfc93b7323f2fb84fe2c9 Mon Sep 17 00:00:00 2001 From: sternenseemann Date: Fri, 31 May 2024 12:09:08 +0200 Subject: [PATCH] haskellPackages.ad: disable problematic test on x86_64-darwin MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit It is somewhat curious that it behaves differently exclusively here, but I don't think it is necessary to stop shipping a package due to floating point arithmetic error—it would be unreasonable to assume there were none… See https://github.com/ekmett/ad/issues/113. --- pkgs/development/haskell-modules/configuration-darwin.nix | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/pkgs/development/haskell-modules/configuration-darwin.nix b/pkgs/development/haskell-modules/configuration-darwin.nix index 701c86200bd5..5f94e9d145c7 100644 --- a/pkgs/development/haskell-modules/configuration-darwin.nix +++ b/pkgs/development/haskell-modules/configuration-darwin.nix @@ -371,4 +371,12 @@ self: super: ({ # same # https://hydra.nixos.org/build/174540882/nixlog/9 jacinda = dontCheck super.jacinda; + + # Greater floating point error on x86_64-darwin (!) for some reason + # https://github.com/ekmett/ad/issues/113 + ad = overrideCabal (drv: { + testFlags = drv.testFlags or [ ] ++ [ + "-p" "!/issue-108/" + ]; + }) super.ad; })