Home | Docs | Forums | Lists | Bugs | Planet | Store | GMN | Get Gentoo!
Not eligible to see or edit group visibility for this bug.
View Bug Activity | Format For Printing | XML | Clone This Bug
Apparently building ghc-6.6.x using ghc-6.4.2 on hppa fails. Awaiting a build log for further analysis.
Created an attachment (id=126459) [details] dev-lang:ghc-6.6:20070730-170815.log ghc-6.6 failing to build ghc-6.6
Created an attachment (id=126461) [details] dev-lang:ghc-6.6.1:20070730-173037.log 6.6 (built by 6.4.2) -> 6.6.1
Created an attachment (id=126525) [details] dev-lang:ghc-6.6.1:20070731-160707.log 6.6.1 (built by 6.4.2) -> 6.6.1
Created an attachment (id=126526) [details] emerge --info
Created an attachment (id=128233) [details] ghc-6.6.1.ebuild The unregisterised build worked fine building ghc-6.6.1 with ghc-6.6.1 (based on ghc-6.4.2) for HPPA (using gcc-4.1.2). The attached ebuild is what I used. It doesn't diff clearly with the ebuild in the tree, so some more work is needed to patch the ebuild in the tree for HPPA support. Based on this build I will be able to build a ghc-bin with minimal sub-arch requirements. Please patch the 6.6.1 ebuild to support HPPA.
Yay! great work Jeroen. So yeah, the ebuild you were working from is more-or-less our current version from our overlay, so what we should do is get that version into portage so you can add the hppa .tbz2 for it.
Created an attachment (id=129070) [details] emerge --info (successful build) I have just built a ghc-bin-6.6.1-hppa.tbz2 with what must be optimal architectural compatibility[1] and I am currently testing it (building haddock, for instance, and later maybe building ghc with it once more). If all goes well, and it finishes building in time, I will upload it to distfiles. [1] CFLAGS='-O2 -pipe -mschedule=700 -march=1.0 -Wall' being the only change to emerge --info.
ghc-bin-6.6.1-hppa.tbz2 is in distfiles-local, ghc-6.6.1.ebuild has the ~hppa keyword and refers to the aforementioned ghc-bin tbz2. *Please* normalise your distribution of ghc now (no regressions, as few as possible dropped keywords) and don't attend to arches piecemeal as you have in the past. Thank you, JeR