diff options
Diffstat (limited to 'sci-libs/qd/files/0009-configure.ac-don-t-set-CC-to-CXX.patch')
-rw-r--r-- | sci-libs/qd/files/0009-configure.ac-don-t-set-CC-to-CXX.patch | 46 |
1 files changed, 46 insertions, 0 deletions
diff --git a/sci-libs/qd/files/0009-configure.ac-don-t-set-CC-to-CXX.patch b/sci-libs/qd/files/0009-configure.ac-don-t-set-CC-to-CXX.patch new file mode 100644 index 000000000000..14269ce2d75a --- /dev/null +++ b/sci-libs/qd/files/0009-configure.ac-don-t-set-CC-to-CXX.patch @@ -0,0 +1,46 @@ +From eff1594d04f140d9e0a4c423a0685481a612c8ca Mon Sep 17 00:00:00 2001 +From: Michael Orlitzky <michael@orlitzky.com> +Date: Wed, 4 Aug 2021 19:15:44 -0400 +Subject: [PATCH 09/12] configure.ac: don't set $CC to $CXX. + +The build system currently sets $CC to whatever $CXX that the +configure script is able to find. This was harmless until now because +C++ is a superset of C, so the C++ compiler should be able to compile +any C code. But in the recently-released autoconf-2.71, the test +program infrastructure no longer emits the + + #ifdef __cplusplus + extern "C" + #endif + +in the same place it used to, which is breaking the test for +name-mangling in the AC_FC_WRAPPERS macro. The macro thinks that it's +compiling C code, but using a compiler that expects C++. While this may +be fixable within autoconf, letting AC_PROG_CC detect the C compiler +separately is both morally correct in this case and happens to solve the +issue anyway. + +This fixes a build issue that was reported on Gentoo linux. + +Gentoo-bug: https://bugs.gentoo.org/775215 +--- + configure.ac | 3 --- + 1 file changed, 3 deletions(-) + +diff --git a/configure.ac b/configure.ac +index 823fa9e..a657e8f 100644 +--- a/configure.ac ++++ b/configure.ac +@@ -84,9 +84,6 @@ case $host in + esac + + AC_PROG_CXX($cxx_list) +-if test "$CC" = ""; then +- CC="$CXX"; +-fi + AC_PROG_CC + + +-- +2.31.1 + |