adaptivecpp: remove wrapper

pkgs/development/compilers/adaptivecpp/default.nix

@@ -80,22 +80,6 @@ stdenv.mkDerivation (finalAttrs: {
   # this hardening option breaks rocm builds
   hardeningDisable = [ "zerocallusedregs" ];
 
-  postFixup =
-    ''
-      wrapProgram $out/bin/syclcc-clang \
-        --prefix PATH : ${
-          lib.makeBinPath [
-            python3
-            lld
-          ]
-        } \
-        --add-flags "-L${llvmPackages.openmp}/lib" \
-        --add-flags "-I${llvmPackages.openmp.dev}/include" \
-    ''
-    + lib.optionalString rocmSupport ''
-      --add-flags "--rocm-device-lib-path=${rocmPackages.rocm-device-libs}/amdgcn/bitcode"
-    '';
-
   passthru = {
     # For tests
     inherit (finalAttrs) nativeBuildInputs buildInputs;

pkgs/development/compilers/adaptivecpp/tests.nix

@@ -2,9 +2,11 @@
   lib,
   stdenv,
   adaptivecpp,
-  # after cursory testing it seems like this isn't really useful, but if errors pop up, maybe look into restricting the targets
-  targetsBuild ? null,
-  targetsRun ? null,
+  # within the nix sandbox, the tests will likely be unable to access the gpu.
+  # for now we just test omp by default as a sanity check,
+  # however the bulk of work in acpp focuses on the generic target, so we want to switch to that.
+  targetsBuild ? "omp",
+  targetsRun ? "omp",
 }:
 stdenv.mkDerivation (finalAttrs: {
   pname = "${adaptivecpp.pname}-tests";