Yup - Jeff found this one this morning. Stupid me testing on x86
only :(. Anyway, this and your other one were just fixed in SVN.
Brian
On Jan 13, 2006, at 9:48 AM, Ralf Wildenhues wrote:
* Ralf Wildenhues wrote on Fri, Jan 13, 2006 at 03:28:09PM CET:
Did you mean this in your recent changes to the trunk?
I'm not quite sure, I don't really understand the code
at a glance.
Hmm, another thing that fails here is that ompi_fortran_integer16_t
gets
set to "not found" on GNU/Linux/x86_64, which causes a compile
failure.
It goes away with the patch below, but that patch is not right,
it just moves the error further to when the preprocessor gets
to mpi.h:
| #if OMPI_HAVE_FORTRAN_INTEGER16
and barfs (while compiling pabort.c):
| ../../../../include/mpi.h:574:32: #if with no expression
| ../../../../include/mpi.h:676:32: #if with no expression
Hope that helps a bit,
Ralf
Index: config/ompi_find_type.m4
===================================================================
--- config/ompi_find_type.m4 (revision 8684)
+++ config/ompi_find_type.m4 (working copy)
@@ -38,7 +38,7 @@
fi
])])
AS_IF([test -z "$oft_real_type"],
- [AS_VAR_SET(type_var, "not found")],
+ [AS_VAR_SET(type_var, "")],
[AS_VAR_SET(type_var, "$oft_real_type")])])
AS_IF([test "AS_VAR_GET(type_var)" == "not found"],
_______________________________________________
devel mailing list
de...@open-mpi.org
http://www.open-mpi.org/mailman/listinfo.cgi/devel
