finalizeGenericPackageDescription: don't rely on defaultComponentRequestedSpec
A record update on defaultComponentRequestedSpec is no safer than just
constructing an instance of the type directly, because there is a mysterious
second constructor, OneComponentRequestedSpec, that our record update won't
match if defaultComponentRequestedSpec turns out to be such a thing for some
reason. Curiously enough, we have helper functions that can query the state of
the ComponentRequestedSpec type, but we cannot easily manipulate it.
commit00e5cd2725182963d1c7b838a1063018fcf3e319
authorPeter Simons <simons@cryp.to>
dateThu 28 Jun 2018, 14:14 (8 years ago)
parent7ccbd66
src/Distribution/Nixpkgs/Haskell/FromCabal.hs+1−1
@@ -53,7 +53,7 @@ finalizeGenericPackageDescription haskellResolver arch compiler flags constraint
5353 finalize resolver = finalizePD flags requestedComponents resolver arch compiler constraints genDesc
5454
5555 requestedComponents :: ComponentRequestedSpec
56− requestedComponents = defaultComponentRequestedSpec
56+ requestedComponents = ComponentRequestedSpec
5757 { testsRequested = True
5858 , benchmarksRequested = True
5959 }