30. toukokuuta 2026 10.22.12 GMT+03:00 Andreas Enge <[email protected]> kirjoitti: >Am Sat, May 30, 2026 at 08:25:15AM +0200 schrieb Andreas Enge: >> that later versions somehow are not available on aarch64 anymore, but >> there have not been changes to ghc-9.4 in this branch. But it could also >> be due to inputs. > >At least the diagnosis is confirmed: >gnu/packages/haskell-xyz.scm:9023:2: warning: package [email protected] does not >support aarch64-linux >pandoc is available on the current master branch. > >The culprit is commit 6749e92d8a85428effa0ccee82df01852a41ce8c : >-;; Versions newer than ghc defined below (i.e. the compiler >-;; haskell-build-system uses) should use ghc-next as their name to >-;; ensure ghc (without version specification) and ghc-* packages are >-;; always compatible. See https://issues.guix.gnu.org/issue/47335. >-(define-public ghc ghc-9.2) >+;; Versions newer than ghc defined below (i.e. the compiler >+;; haskell-build-system uses) should use ghc-next as their name to >+;; ensure ghc (without version specification) and ghc-* packages are >+;; always compatible. See https://issues.guix.gnu.org/issue/47335. >+(define-public ghc ghc-9.10) > >But already ghc-9.4 does not build on aarch64. This is a serious >regression, and I think we cannot merge like this.
Agreed. >Would it be enough to drop this commit for now? That is a necessary commit for the ones after it, which are the point of this branch. >Longer term, what would it take to build newer haskell versions >on aarch64? I will investigate that tommorrow and hope to get it fixed in a few days. - Saku
