[kepler-dev] composite actor library

[Wibke Sudholt]

Hi Chad,

Ilkay will look into this tonight (later today on your side of the
Atlantic), she has no internet connection at the moment. The behavior you
describe is what we tried to explain before in bug #1851. We thought this is
related to the way how Kepler clones the actors for the actor search
interface.

Best regards,

Wibke


On 12.04.2005 19:58, "Chad Berkley" <berkley at nceas.ucsb.edu> wrote:

> Hey Ilkay and Wibke,
> 
> Several of us are seeing errors with various things that seem to all be
> stemming from either work you are doing on the composite actors or work
> i'm doing on the moml parser.  It seems that the work you are doing does
> not work with the CVS version of PTII.  The work i'm doing doesn't work
> with the release version of PTII (because they've changed the api for
> the moml parser).  But even when we run kepler with the release version
> of PTII, dan, jing and I are seeing an error whenever we try to drag a
> composite from the library to the canvas that informs us that a port
> cannot be link to a relation inside a composite.  Have you seen this
> error?
> 
> When I try to run your stuff with the cvs version, I see a bunch of
> errors that says some of your code is using the wrong constructor format
> for creating ports:
> 
> Caused by: ptolemy.kernel.util.IllegalActionException: Could not find a
> port named 'result' or 'resultPort'. This can occur when the name of the
> variable does not match the name passed to the constructor of the actor.
>     [java] Right:
>     [java]     result = new TypedIOPort(this, "result", true, false);
>     [java] Right:
>     [java]     result = new TypedIOPort(this, "resultPort", true, false);
>     [java] Wrong:
>     [java]     result = new TypedIOPort(this, "foo", true, false);
>     [java]   in .configuration.actor library.String Concatenator
>     [java]     at ptolemy.kernel.Entity.clone(Entity.java:199)
>     [java]     ... 52 more
> 
> Could you take a look and try to see if the error is easily fixable?  I
> think there will be some serious implications of writing code that
> depends on the 4.0 api instead of the newest api.  We had major problems
> the last time that ptolemy went from 3 to 4, so I'm trying to develop to
> their newest API whenever possible.  If you see a problem with the
> approach of developing to the CVS version of ptolemy, let me know.
> 
> chad