[kepler-users] Branching a workflow!

Bertram Ludaescher ludaesch at ucdavis.edu
Thu Jul 17 17:39:24 PDT 2008


On Wed, Jul 16, 2008 at 5:30 AM, Edward A. Lee <eal at eecs.berkeley.edu>
wrote:

>
> This sounds somewhat similar to the way Case works:
>
> Case is multi-composite actor (it has any number of
> models inside it). The models inside are presented in a
> tabbed pane (I guess this is why it's not in Kepler).
> It has a control input port that specifies which of
> the inside models to execute. The control can be
> any data type, and the inside model's name (and tab
> in the tabbed pane) is simply the string representation
> of the value of the control.
>
> IMHO: Using BooleanSwitch and BooleanSelect for control
> flow is a bit like using goto in an imperative language.
> It doesn't scale well...
>

I couldn't agree more. We've found workflows with lots of those control-flow
components to be hard to digest, comprehend, evolve. Higher-order actors and
related approaches seem to work much better there.

Bertram



>
> Edward
>
> At 12:43 AM 7/16/2008, Bertram Ludaescher wrote:
> >One more note: Soon, a number of so-called "COMAD extensions" should be
> available from the Kepler repository. In this workflow design model, data
> dependent routing and conditional execution can be handled differently
> >(and arguably at a higher level of abstraction).
> >
> >The basic idea is that channels transport data collections (think: XML
> data) and actors can be configured to work only on certain parts of those
> data collections.
> >In such a setting data dependent routing is no longer a user-concern (or
> much less so; a dataflow optimizing compiler takes care of this instead).
> >Similarly, conditional execution can be replaced by conditional "match
> rules" of COMAD actors.
> >
> >We'll add some references on the underlying model to the kepler web site.
> The Kepler COMAD extensions should become available over the next couple of
> weeks as well (COMAD gurus, correct me if I'm wrong ;-)
> >
> >Bertram
> >
> >On Tue, Jul 15, 2008 at 12:47 PM, Dan Higgins <<mailto:
> higgins at nceas.ucsb.edu>higgins at nceas.ucsb.edu> wrote:
> >Unless someone has made some changes in the past few months, the Case
> actor does not work in Kepler due to changes Kepler has made in the way the
> workflow GUI is displayed (related to the actor and search widgets on the
> left?)
> >
> >I agree - the Case actor should appear in Kepler!!!  If there is not a bug
> about it, someone should add one.
> >
> >Dan Higgins
> >
> >=====
> >
> >
> >Edward A. Lee wrote:
> >How about using the Case actor? It can be used with SDF.
> >It's in the Ptolemy library under "Higher-Order Components".
> >Not sure where (whether) it's in Kepler...  If it isn't, it
> >should be... It should also be highlighted on the "how to"
> >web page...
> >
> >Edward
> >
> >
> >At 10:28 AM 7/14/2008, Josep Maria Campanera Alsina wrote:
> >
> >Dear All,
> >I'd like to pick some ideas from audience:
> >
> >I'd like to branch my workflow! Actually I'd like to be able to decide to
> run some parts of my workflow while keep some others off! How can I do that
> in the SDF paradigm? I thought to use some actors from "Workflow control"
> and "Boolean control"?
> >For instance If I set up a parameter (YES/NO) and according to that, some
> parts of the workflow can execute (YES) and some have not (NO). How to do
> that?
> >
> >Thanks a lot,
> >
> >Campa,
> >_______________________________________________
> >Kepler-users mailing list
> ><mailto:Kepler-users at ecoinformatics.org>Kepler-users at ecoinformatics.org
> >
> http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-users
> >
> >
> >
> >------------ Edward A. Lee
> >Robert S. Pepper Distinguished Professor
> >EECS Dept., 545Q Cory Hall, UC Berkeley, Berkeley, CA 94720-1770
> >phone: 510-642-0253, fax: 510-642-2845
> ><mailto:eal at eecs.Berkeley.EDU>eal at eecs.Berkeley.EDU,
> http://www.eecs.berkeley.edu/Faculty/Homepages/lee.html
> >_______________________________________________
> >Kepler-users mailing list
> ><mailto:Kepler-users at ecoinformatics.org>Kepler-users at ecoinformatics.org
> >
> http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-users
> >
> >
> >
> >
> >_______________________________________________
> >Kepler-users mailing list
> ><mailto:Kepler-users at ecoinformatics.org>Kepler-users at ecoinformatics.org
> >
> http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/kepler-users
> >
>
> ------------
> Edward A. Lee
> Robert S. Pepper Distinguished Professor
> EECS Dept., 545Q Cory Hall, UC Berkeley, Berkeley, CA 94720-1770
> phone: 510-642-0253, fax: 510-642-2845
> eal at eecs.Berkeley.EDU,
> http://www.eecs.berkeley.edu/Faculty/Homepages/lee.html
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mercury.nceas.ucsb.edu/ecoinformatics/pipermail/kepler-users/attachments/20080717/164897f9/attachment.html>


More information about the Kepler-users mailing list