[mpich-discuss] Building with Clang

[Dave Goodell]

On Aug 20, 2011, at 10:00 PM CDT, Jed Brown wrote:

> On Mon, Aug 15, 2011 at 15:46, Dave Goodell <goodell at mcs.anl.gov> wrote:
> Can you send us the c.txt and config.log files indicated in the README?  Perhaps there's something that you missed that one of us will spot.  Also, be sure to let us know exactly which version of the code that those logs are coming from.
> 
> Just updated to r8940, c.txt and config.log attached.
> <c.txt><config.log>

The immediate cause of the configure failure can be seen in the config.log.  For some reason the configure test is generating a C program that looks like this:

----8<----
void main(){ return 0; }
----8<----

Which clang reasonably complains about:

----8<----
conftest.c:11:9: error: 'main' must return 'int'
        void main(){ return 0; }
        ^
conftest.c:11:22: error: void function 'main' should not return a value [-Wreturn-type]
        void main(){ return 0; }
                     ^      ~
----8<----

However, it's not obvious to me why that code is getting generated by configure.  When I use autoconf 2.68 (as you do) and try a few simple autoconf macros that imitate what MPICH2 is doing, I get a well-formed conftest.c:

----8<----
int
main ()
{

  ;
  return 0;
}
----8<----

I'll do a bit more testing here with actual MPICH2 to see if I can reproduce what you're seeing.    MPICH2 is definitely using AC_LINK_IFELSE here in an unconventional way, so it would not be surprising to see that autoconf-2.68 somehow breaks the shared library support.

In the meantime can you send me your actual raw configure script as well?  That might shed some light on what's happening here.

-Dave