diff options
| author | Mitch Tishmack <mitch.tishmack@gmail.com> | 2017-07-30 11:39:11 -0500 |
|---|---|---|
| committer | Natanael Copa <ncopa@alpinelinux.org> | 2018-06-12 07:36:35 +0000 |
| commit | 0194bb7bf46305018dcebec6f44468c28da2c401 (patch) | |
| tree | f28364d33aff2c323a02aab15c4d38ba15c55827 /community/ghc/0002-Correct-issue-with-libffi-and-glibc.patch | |
| parent | 32571b1d45067ededceb0ca113967b9d9ee0ab37 (diff) | |
| download | aports-0194bb7bf46305018dcebec6f44468c28da2c401.tar.bz2 aports-0194bb7bf46305018dcebec6f44468c28da2c401.tar.xz | |
community/ghc: upgrade to 8.4.2
There is only one package that depends on ghc right now, and that is
community/cabal.
Removed armhf support for now as there appear to be overall issues with
that architecture across libc's. Additionally, for 8.6 to build we will
need to update to at least 8.2. As such, dropping support for armhf
makes the most sense for right now. I'll continue working with upstream
to fix it.
Removed the haskell-* provides as that was originally a plan to follow
what arch was doing for their haskell setup. In lieu of that not
providing value removing them entirely.
The ghc test suite was updated to use python 3 from python 2.
Finally really late in the release candidate cycle a change to ld
detection necessitated the need to pass --disable-ld-override to the
configure of ghc.
Without that in place there appears to be some shenanigans going on
in exported symbols.
Ghc upstream ticket:
https://ghc.haskell.org/trac/ghc/ticket/13958
Offending commmit:
https://git.haskell.org/ghc.git/commitdiff/2785ef0e31a123400da950ffafebe6cb1ce3f4eb
Diffstat (limited to 'community/ghc/0002-Correct-issue-with-libffi-and-glibc.patch')
| -rw-r--r-- | community/ghc/0002-Correct-issue-with-libffi-and-glibc.patch | 28 |
1 files changed, 0 insertions, 28 deletions
diff --git a/community/ghc/0002-Correct-issue-with-libffi-and-glibc.patch b/community/ghc/0002-Correct-issue-with-libffi-and-glibc.patch deleted file mode 100644 index 87420f9928..0000000000 --- a/community/ghc/0002-Correct-issue-with-libffi-and-glibc.patch +++ /dev/null @@ -1,28 +0,0 @@ -From 843bd68ea83ba011d593ad7162d1df89c230004b Mon Sep 17 00:00:00 2001 -From: Mitch Tishmack <mitch.tishmack@gmail.com> -Date: Sat, 15 Oct 2016 21:53:13 -0500 -Subject: [PATCH 1/1] Correct issue with libffi and glibc - -Ref: https://github.com/mitchty/alpine-linux-ghc-bootstrap/issues/16 - -Shamelessly learnt from here -https://github.com/nilcons/ghc-musl/blob/master/ghc-musl/Dockerfile#L24 ---- - libffi/ghc.mk | 1 + - 1 file changed, 1 insertion(+) - -diff --git a/libffi/ghc.mk b/libffi/ghc.mk -index 404cce9..2034873 100644 ---- a/libffi/ghc.mk -+++ b/libffi/ghc.mk -@@ -61,6 +61,7 @@ $(libffi_STAMP_CONFIGURE): $(TOUCH_DEP) - # We have to fake a non-working ln for configure, so that the fallback - # option (cp -p) gets used instead. Otherwise the libffi build system - # will use cygwin symbolic links which cannot be read by mingw gcc. -+ sed -i s/__gnu_linux__/1/ libffi/build/src/closures.c - chmod +x libffi/ln - - # We need to use -MMD rather than -MD, as otherwise we get paths --- -2.9.0 - |