There is only one system call in CitcomCU, fixed in r8252.

All remaining system calls in CitcomS are in lib/Output_gzdir.c, which is for
compressed ASCII output. Most users will use normal ASCII output and will not be
affected by this issue.

lib/Instructions.c:1429:      sprintf(message,"rm %s",files); system(message);
lib/Pan_problem_misc_functions.c:83:    system(command); /* copy */

These two are fixed in r8244.

Shijie reported that the mpi implementation on Teragrid TACC doesn't support
system call. 

The list of system call in CitcomS trunk:

lib/Instructions.c:1429:      sprintf(message,"rm %s",files); system(message);
lib/Output_gzdir.c:501:      system(output_file2); /* copy */
lib/Output_gzdir.c:1246:      system(mstring);  /* unzip */
lib/Output_gzdir.c:1274:  system(command_string); /* zip */
lib/Pan_problem_misc_functions.c:83:    system(command); /* copy */


=====================================
Hi Eh,

Have anyone tried citcomS on lonestar? 

The following is what I got from Luke Wilson from TACC. He basically said that
we cannot have any system calls. He identified one in the code, and suggested
that it be removed. 
Does anyone of you would like to give it try?

====================================================
Shijie,

There appears to be a system() call in your code in file
Pan_problem_misc_functions.c:

sprintf(command, "cp -f %s %s\n", name, unique_name);
system(command);

The error with your code:
Abort: [c41-305:18] Got completion with error, code=12
 at line 2352 in file viacheck.c

Is consistent with the behavior we record when a system call has been made in
MVAPICH. If possible, try rewriting this line using standard C library calls
instead of a system() call and see if that resolves the error.

Lucas Wilson
==================================================

Please let me know how it goes.


Cheers,

Shijie