[kepler-dev] Running Shell Scripts inside Kepler

[Christopher Brooks]

BTW - My hack for the exec command in ptjacl (a 100% Java
implementation of a subset of Tcl)
was to add a -sh command line argument.  If -sh is present, then
we use /bin/sh, if not, then we use cmd under Windows etc.

The ptjacl exec command is written in Java and uses the
Runtime.exec(), so it is very similar to CommandLine.

I'm not sure if this would help with CommandLine.  It seems dumb
to have to tell the tool whether to use /bin/sh or not, but at
the time it seemed like the quickest solution.  It looks
like the problem for here is that sometimes you want /bin/sh,
sometimes you don't.  I'm surprised that under Linux and
the Mac, the following did not work for you.

              commandArray[0] = "/bin/sh";
              commandArray[1] = "-c";
              commandArray[2] = "R";
              commandArray[3] = "--silent";
              commandArray[4] = "--no-restore";
              commandArray[5] = "--no-save";

BTW - Under Windows the ptjacl exec command writes out two files
c:/temp/jacl1.bat and c:/temp/jacl2.bat

jacl1.bat looks like:
--start--
@echo off
cmd.exe /C C:\TEMP\jacl2.bat
--end--

--start--
@echo off
C:
cd \cxh\ptII\ptolemy\copernicus\java\test
make "-C" "../cg/ComplexDivide" "MODEL=ComplexDivide"
"SOURCECLASS=../../../../ptolemy/actor/lib/test/auto/ComplexDivide.xml"
"smokeTest" 
--end--

This seems a litle byzantine, but it works.

_Christopher



--------

    Hi Ilkay (and others)
    
    A few weeks ago, you were asking if I had experience/problems with 
    running shell scripts using the CommandLine actor. At the time, I did 
    not, but that situation changed with some recent work on the RExpression 
    actor. And I discovered some problems (and solutions) which may be of 
    interest.
    
    In the commandLine actor (and Java in general), the runtime.exec() 
    method is used to create subprocesses; e.g.
    
              _process = runtime.exec(commandArray);
    
    where, for the Windows OS, the commandArray might be something like
    
              commandArray[0] = "cmd.exe";
              commandArray[1] = "/C";
              commandArray[2] = "R";
              commandArray[3] = "--silent";
              commandArray[4] = "--no-restore";
              commandArray[5] = "--no-save";
p    
    In this case, the Windows command line processor 'cmd.exe' is used to 
    execute the 'R' program. This works fine on a Windows machine and use of 
    'cmd' to launch the R program means that the PATH system variable is 
    used to find R. Thus no need for specifying the absolute path.
    
    Now, on Linux or MacOSX, the commandArray (for the CommandLine actor) 
    uses /bin/sh to execute the command line; e.g.
    
              commandArray[0] = "/bin/sh";
              commandArray[1] = "-c";
              commandArray[2] = "R";
              commandArray[3] = "--silent";
              commandArray[4] = "--no-restore";
              commandArray[5] = "--no-save";
    
    Now here is where I ran into trouble!!! On Linux and the Mac, 'R' is a 
    shell script, not a simple executable! And apparently, the Java 
    runtime.exec() method WILL NOT START A SUBPROCESS SET UP THIS WAY WITH A 
    SHELL SCRIPT!!!
    
    So what I discovered was that on Unix machines, I can change the 
    commandArray to directly call the shell script (i.e. not call the shell 
    to launch it). In this case, the commandArray is just
    
              commandArray[0] = "R";
              commandArray[1] = "--silent";
              commandArray[2] = "--no-restore";
              commandArray[3] = "--no-save";
    
    This worked on both Linux and the MacOSX and the OS still can find a 
    shell script using PATH (which is cannot do under Windows).
    
    So this solved my problem, but I am not sure exactly how it might work 
    for the CommandLine actor where it is not obvious whether or not one is 
    trying to run a shell script. Maybe, we can just drop the '/bin/sh' part 
    of the commandArray for executables AND shell scripts?
    
    Dan
    
    -- 
    *******************************************************************
    Dan Higgins                                  higgins at nceas.ucsb.edu
    http://www.nceas.ucsb.edu/    Ph: 805-893-5127
    National Center for Ecological Analysis and Synthesis (NCEAS) 
    Marine Science Building - Room 3405
    Santa Barbara, CA 93195
    *******************************************************************
    
    _______________________________________________
    Kepler-dev mailing list
    Kepler-dev at ecoinformatics.org
    http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-dev
--------