svn commit: r541283 - head/net/mpich/files
Tobias Kortkamp
tobik at FreeBSD.org
Sun Jul 5 14:55:09 UTC 2020
Author: tobik
Date: Sun Jul 5 14:55:08 2020
New Revision: 541283
URL: https://svnweb.freebsd.org/changeset/ports/541283
Log:
net/mpich: Unbreak build with GCC 10
checking whether gfortran10 allows mismatched arguments... no
configure: error: The Fortran compiler gfortran10 will not compile files that call
the same routine with arguments of different types.
http://package18.nyi.freebsd.org/data/121amd64-default-PR244494/2020-06-10_18h04m44s/logs/errors/mpich-3.2.1_6.log
PR: 246700
Added:
head/net/mpich/files/patch-configure (contents, props changed)
head/net/mpich/files/patch-test_mpi_configure (contents, props changed)
Added: head/net/mpich/files/patch-configure
==============================================================================
--- /dev/null 00:00:00 1970 (empty, because file is newly added)
+++ head/net/mpich/files/patch-configure Sun Jul 5 14:55:08 2020 (r541283)
@@ -0,0 +1,30 @@
+--- configure.orig 2017-11-11 02:22:24 UTC
++++ configure
+@@ -32904,7 +32904,7 @@ rm -f core conftest.err conftest.$ac_objext conftest.$
+ # The best solution is to turn off errors on particular routines
+ # if that isn't possible (e.g., too many of them), then
+ # just try arguments that turn off all checking
+- for flags in "-mismatch" ; do
++ for flags in "-fallow-argument-mismatch" "-mismatch" ; do
+ testok=no
+ FFLAGS="$FFLAGS $flags"
+ cat > conftest.$ac_ext <<_ACEOF
+@@ -32946,17 +32946,7 @@ the same routine with arguments of different types." "
+ fi
+
+ addarg=$pac_cv_prog_f77_mismatched_args_parm
+-
+- if test "X$addarg" != "X" ; then
+- # We could add the names of all of the MPI routines that
+- # accept different types. Instead, we fail cleanly.
+- # Some Fortran compilers allow you to turn off checking for
+- # mismatched arguments for *all* routines. Adding an argument
+- # that turns off checking for *everything* is not something that
+- # configure should do - if the user wants this, they can follow
+- # the instructions in the following error message.
+- as_fn_error $? "The Fortran compiler $F77 does not accept programs that call the same routine with arguments of different types without the option $addarg. Rerun configure with FFLAGS=$addarg" "$LINENO" 5
+- fi
++FFLAGS="$FFLAGS $addarg"
+
+ bindings="$bindings f77"
+
Added: head/net/mpich/files/patch-test_mpi_configure
==============================================================================
--- /dev/null 00:00:00 1970 (empty, because file is newly added)
+++ head/net/mpich/files/patch-test_mpi_configure Sun Jul 5 14:55:08 2020 (r541283)
@@ -0,0 +1,30 @@
+--- test/mpi/configure.orig 2020-07-05 14:40:24 UTC
++++ test/mpi/configure
+@@ -9726,7 +9726,7 @@ rm -f core conftest.err conftest.$ac_objext conftest.$
+ # The best solution is to turn off errors on particular routines
+ # if that isn't possible (e.g., too many of them), then
+ # just try arguments that turn off all checking
+- for flags in "-mismatch" ; do
++ for flags in "-fallow-argument-mismatch" "-mismatch" ; do
+ testok=no
+ FFLAGS="$FFLAGS $flags"
+ cat > conftest.$ac_ext <<_ACEOF
+@@ -9768,17 +9768,7 @@ the same routine with arguments of different types." "
+ fi
+
+ addarg=$pac_cv_prog_f77_mismatched_args_parm
+-
+- if test "X$addarg" != "X" ; then
+- # We could add the names of all of the MPI routines that
+- # accept different types. Instead, we fail cleanly.
+- # Some Fortran compilers allow you to turn off checking for
+- # mismatched arguments for *all* routines. Adding an argument
+- # that turns off checking for *everything* is not something that
+- # configure should do - if the user wants this, they can follow
+- # the instructions in the following error message.
+- as_fn_error $? "The Fortran compiler $F77 does not accept programs that call the same routine with arguments of different types without the option $addarg. Rerun configure with FFLAGS=$addarg" "$LINENO" 5
+- fi
++FFLAGS="$FFLAGS $addarg"
+
+ # Check whether we need -lU77 to get iargc and getarg, which
+ # are used for a few of the tests in spawn (U77 was needed with
More information about the svn-ports-all
mailing list