[Esml-users] Re: RE Esml-users Event ports with null event refs

[Schulte, Mark D]

Right.  The INTERVAL_TIMEOUT inconsistencies arose from the fact that as
I described, we are populating that ACL data from information that
people are supposed to be capturing in the documentation requirements
for the classes.  Given that, it is hard to remove the inconsistencies
and misspellings between different developers, particularly because when
the documentation standard was originally set, no one envisioned using
automatic tools to do something with this information.  But we should go
through the model and make everything INTERVAL_TIMEOUT.  Dennis could
explain this further, because I think the goal would be to support more
event types in the future, which we've accounted for in ESML.  But for
right now, Sandeep is right in that DATA_AVAILABLE merely captures all
the other non-timeout events in our system.  So strictly speaking, there
really isn't any event type checking going on right now.   

-----Original Message-----
From: Sandeep Neema [mailto:sandeep at isis.vanderbilt.edu] 
Sent: Thursday, May 13, 2004 11:41 AM
To: esml-users
Cc: Noll, Dennis D; Schulte, Mark D
Subject: RE: [Esml-users] Re: RE Esml-users Event ports with null event
refs


Jesse,
Yes, the 'Interval Timeout', and 'INTERVAL_TIMEOUT', and so on should be
the same. These are user modeling errors i.e. when somebody is editing
the ACL properties the mistakes are made. However, the 'ANY' simply
indicates that the component can publish any event that it is configured
to publish during thru the models. There is nothing magical about the
DATA_AVAILABLE event as far as the component types are concerned, and
that's why it is not there in the component type models. The
DATA_AVAILABLE event is a scenario specific artifact that is modeled in
the scenario models, and the component instances are appropriately bound
to that particular event type. The 'INTERVAL_TIMEOUT' on the other hand
has special semantics and is an event that could/should be published by
the Timer component (which is not a real component) only.

In any case these questions are best answered by the Boeing folks (I am
cc'ing Dennis and Mark on this)

Thanks,
Sandeep

--
Sandeep Neema
Research Scientist,
Institute of Software Integrated Systems, Vanderbilt University
Phone: 615-343-9996
Email: sandeep.k.neema at vanderbilt.edu

-----Original Message-----
From: Jesse Greenwald [mailto:jesse at cis.ksu.edu] 
Sent: Thursday, May 13, 2004 11:23 AM
To: esml-users
Subject: RE: [Esml-users] Re: RE Esml-users Event ports with null event
refs

Hi Sandeep,

After looking at OEP Rose model and the XML ESCM export, I think there
are some problems with the model.  For instance, there three different
event types for timeouts (Interval Timeout, INTERVAL_TIMEOUT,
INTERVAL_TIMEOUT).  Shouldn't these all be the same?  There should also
be an event type for data available events since this is the other event
type that is used in the system.  Instead, there is no data available
event and components are being specified as just publishing ANY.  But,
this would seem to imply that the components are publishing timeout
events, but only timers should be publishing timeout events.

I'm going to go ahead and ping the Boeing folk and see what they say
about the model.

Thanks a bunch for help,
Jesse

On Wed, 2004-05-12 at 13:51, Sandeep Neema wrote:
> Jesse,
> These are the ACL properties that I am referring to. I believe that
you
> need the Teknowledge plug-in to look at/or edit those attributes in 
> Rose. The Teknowlege plug-in generates the PRISM xml file that is used

> to seed the ESML model.
> 
> Here is an XML generated with that plugin.
> 
> Thanks
> Sandeep
> 
> 
> --
> Sandeep Neema
> Research Scientist,
> Institute of Software Integrated Systems, Vanderbilt University
> Phone: 615-343-9996
> Email: sandeep.k.neema at vanderbilt.edu
> 
> -----Original Message-----
> From: Jesse Greenwald [mailto:jesse at cis.ksu.edu]
> Sent: Wednesday, May 12, 2004 12:32 PM
> To: esml-users
> Subject: RE: [Esml-users] Re: RE Esml-users Event ports with null
event
> refs
> 
> Hi Sandeep,
> 
> On Tue, 2004-05-11 at 22:57, Sandeep Neema wrote:
> > Jesse,
> > The specific events to publish are often instance-specific decision
> i.e.
> > a Component user can customize the specific event-types published by
a
> > component when instantiating a component in the system. This is the 
> > reason why these event references are 'null'. In fact this
information
> > (i.e. null event-types) is coming from the PRISM interface of
Rational
> > Rose, where some publish/subscribe ports of components are typed
with
> > 'ANY Event'.
> 
> Is the Rational Rose model that you are talking about the one that is 
> included with the OEP releases from Boeing?  I've looked at the Rose 
> model that comes with the 3.0 release of the OEP and I don't see what 
> you are referring too.  In the model I'm look at, most of the
components
> simply use the abstract BM__EventSink (of which there is only one 
> implementation - BM__UpdateEventSinkAdapter_T) for consuming events
and
> the UUEventSupplier for publishing events.  Is there some attribute or

> association that I'm missing?
> 
> Thanks,
> Jesse
> 
> _______________________________________________
> esml-users mailing list
> esml-users at list.isis.vanderbilt.edu
> http://list.isis.vanderbilt.edu/mailman/listinfo/esml-users
> 
> ______________________________________________________________________
> _______________________________________________
> esml-users mailing list
> esml-users at list.isis.vanderbilt.edu
> http://list.isis.vanderbilt.edu/mailman/listinfo/esml-users

_______________________________________________
esml-users mailing list
esml-users at list.isis.vanderbilt.edu
http://list.isis.vanderbilt.edu/mailman/listinfo/esml-users