[kepler-dev] Nondeterministic merge in PN domain

Christopher Brooks cxh at eecs.berkeley.edu
Fri Jul 14 16:27:02 PDT 2006

Hi Ilkay,

Just yesterday, Antoon Goderis, Edward and I were looking over the
Promoter Identification Workflow (PIW) workflow from
https://www-casc.llnl.gov/sdm/download.php that Xiaowen created.

The PIW workflow is available via Webstart:

That workflow is uses PN and basically takes advantage
of a bug in older versions of PN where non-deterministic
merges were easy to make.  Now, to make a non-deterministic
merge, one must use the NonDeterministicMerge actor.
Where we saw this was in the Setup composite actor.
The Setup is a little wacky, there is probably a better way.
Edward might have more insight on this.

The version of the PIW demo from webstart fails with:
ptolemy.kernel.util.IllegalActionException: The length of the
specified array is zero.
	  at ptolemy.data.ArrayToken._initialize(ArrayToken.java:575)
	  at ptolemy.data.ArrayToken.<init>(ArrayToken.java:68)
	  at ptolemy.domains.sdf.lib.SequenceToArray.fire(SequenceToArray.java:139)
	  at ptolemy.actor.process.ProcessThread.run(ProcessThread.java:181)

The version of the PIW demo in Kepler 
 (Start Kepler, Help -> Documentation -> Programmer Documentation
  -> Promotor Identification Workflow)
fails with
java.lang.OutOfMemoryError: unable to create new native thread
    at java.lang.Thread.start(Native Method)
    at org.apache.axis.wsdl.gen.Parser.run(Parser.java:306)
    at org.sdm.spa.WebService.fire(WebService.java:727)
    at ptolemy.actor.process.ProcessThread.run(ProcessThread.java:186)

BTW - We are working on a short paper that discusses embedding domains
and Antoon proposed working with the PIW example.  If
you can get it working under PN, then I'd like to see about converting
it to DDF or ideally SDF.



    Hi Christopher,
    The workflows don't use the NondeterministicMerge actor but implement  
    nondeterministic merge in the workflow.
    Until PTII 5.x, the PN director allowed this, but I'm trying to  
    figure out why the workflow was implemented that way, and the  
    workflow developer (Xiaowen--she left the project) insisted on  
    keeping the old director. It was to avoid a race condition. I'll try  
    to port the workflow to the current PTII version if I understand what  
    the problem with using the new version of the PNDirector is.
    On Jun 22, 2006, at 3:23 PM, Christopher Brooks wrote:
    > Hi Ilkay,
    > I'm not sure.  Edward checked in a working nondeterministic merge
    > on 03 Oct 2004.
    > I'm a little confused though, because the CVS head of Ptolemy has a
    > working nondeterministic merge actor (pn.kernel.NondeterministicMerge)
    > and a PN demo: pn/demo/NondeterministicMerge/NondeterministicMerge.xml
    > _Christopher
    > --------
    >     Hi,
    >     Does anyone know what exactly have changed between the version  
    > of the
    >     PNDirector that accepts nondeterministic merge and the current  
    > one?
    >     There was an email thread on this but we couldn't find it.
    >     Nandita and I are trying to port older workflows that use
    >     nondeterministic merge under PN domain to the newer version of PN.
    >     Thanks!
    >     -ilkay
    > ---------------------------------------------------------------------- 
    > -----
    >    -
    >     Posted to the ptolemy-hackers mailing list.  Please send  
    > administrative
    >     mail for this list to: ptolemy-hackers- 
    > request at ptolemy.eecs.berkeley.edu
    > --------
    Ilkay ALTINTAS
    Assistant Director, National Laboratory for Advanced Data Research  
    Manager, Scientific Workflow Automation Technologies (SWAT) Lab
    San Diego Supercomputer Center(SDSC), UCSD
    9500 Gilman Drive, MC: 0505  La Jolla, CA  92093-0505
    phone: (858) 822-5453                        fax: (858) 822-3693
                     web: http://users.sdsc.edu/~altintas
    Posted to the ptolemy-hackers mailing list.  Please send administrative
    mail for this list to: ptolemy-hackers-request at ptolemy.eecs.berkeley.edu

More information about the Kepler-dev mailing list