[gme-users] Intersection of aspects the referred modelsmighthaveis empty?

[Ace Thompson]

Matt,

 

Thanks for the explanation on this, especially the part that implies that
running into this situation doesn't mean that my language is all screwed up
:-)

 

If it's not already there, just a $.02 suggestion, it might be worth
mentioning something about this topic in the GME on-line documentation in
one of the sections describing Aspects. I think I've run into it twice now
without even trying, but then again sometimes I have bad luck with these
things . [you'd think I would've remembered straight off how to deal with it
the 2nd time, right? :-) ]

 

Along those lines, has there been any talk or consideration of a Wiki-like
forum for GME users?  I'm aware of the JIRA tool but that seems more focused
on GME bugs/issues, etc.  A Wiki kind of tool might make a nice complement
to the gme-users list. [with enough GME end-user contributions, of course
:-) ]   As an added bonus, Wikis aren't out of style yet so it'd be totally
trendy! 

 

Regards,

-ace

 

  _____  

From: gme-users-bounces at list.isis.vanderbilt.edu
[mailto:gme-users-bounces at list.isis.vanderbilt.edu] On Behalf Of Matthew J.
Emerson
Sent: Monday, June 05, 2006 3:09 PM
To: gme-users
Subject: RE: [gme-users] Intersection of aspects the referred
modelsmighthaveis empty?

 

The error is a limitation imposed by the way GME visualizes references to
Models that expose ports.

 

In a particular Model, one Aspect may include one set of ports object and a
second Aspect may include a second set of port objects. GME has to know
which set of ports to show in every different place where that kind of Model
can be created - basically, which of the Aspect of the Model gives its
external interface for a particular context.

 

The same thing holds true for References to Models. However, one Reference
can refer to two different Model types. If the different Model types to
which a Reference may refer have no common set of Aspects, then GME might
not be able to visualize a consistent interface for objects of that
Reference type.

 

GME chooses the maximally-restrictive solution for this problem by requiring
that the set of models that may be referred to be a Reference object share
at least one Aspect. This lets GME visualize each reference in each context
the same way regardless of what the referent object type is.

 

I think this is a technical/implementation issue and doesn't say much about
language design.

 

"does the fact that the aspect is shared between the two models have any
significance?"

 

Not in GME (currently). Aspects are purely a visualization/complexity
management concern. It happens that you can only draw a line between two
things that show up on the screen at the same time, so related objects end
up in the same Aspect. That doesn't mean that all of the objects in the same
Aspect belong to any one conceptually-distinct important facet of their
containing object - this is why I often have trouble thinking up good Aspect
names.

 

Similarly, imagine that case that you want to have a Reference which refers
to two different Model types, neither of which exposes any ports in any
Aspect. Does it really matter if they don't share a common Aspect, when they
will be visualized the same regardless? There's no harm in introducing
"dummy" aspects in such a case to get a paradigm to pass muster - in such a
case GME's implementation is just overly-restrictive.

 

--Matt

 

  _____  

From: gme-users-bounces at list.isis.vanderbilt.edu
[mailto:gme-users-bounces at list.isis.vanderbilt.edu] On Behalf Of Ace
Thompson
Sent: Monday, June 05, 2006 3:53 PM
To: gme-users
Subject: RE: [gme-users] Intersection of aspects the referred models
mighthaveis empty?

 

Sandeep,

 

Thanks, this is definitely the situation I have.  Basically my "E" element
(an FCO) has several inheritors, one of which is a <<Model>> and another one
is a <<Reference>> to a Model.  The second model has an aspect "A"; when I
add this aspect to the first model, it eliminates the error.  This isn't
really what I want language-wise, it's sort of spurious, but it seems to get
around the issue until I have time to sit down and figure out if I've made
some fundamental language-design no-no on the way to getting myself into
this situation :-) 

 

So while we're on the topic, what does it mean for two different models to
have the same <<Aspect>>, if the two different models aren't in any other
way related and don't contain any common sub-elements?   GME and the
Meta-interpreter don't seem to mind when I do this, and I get what I would
expect to get in each of the models if I had only assigned that Aspect to
that one model, but I guess the question is: does the fact that the aspect
is shared between the two models have any significance?

 

Thanks a lot,

-ace

 

  _____  

From: gme-users-bounces at list.isis.vanderbilt.edu
[mailto:gme-users-bounces at list.isis.vanderbilt.edu] On Behalf Of Sandeep
Neema
Sent: Monday, June 05, 2006 11:54 AM
To: gme-users
Subject: RE: [gme-users] Intersection of aspects the referred models
mighthave is empty?

 

Ace,

I don't exactly recall it either, but I think this may be related to when
you have a reference to a base class and there are multiple derived classes
with conflicting aspects.

 

HTH,

Sandeep

 

  _____  

From: gme-users-bounces at list.isis.vanderbilt.edu
[mailto:gme-users-bounces at list.isis.vanderbilt.edu] On Behalf Of Ace
Thompson
Sent: Monday, June 05, 2006 1:44 PM
To: gme-users
Subject: [gme-users] Intersection of aspects the referred models might have
is empty?

 

Hi,

 

I have a metamodel containing a <<Reference>> "E" and when I try to register
with the MetaGME Interpreter I get the following error in the
Metainterpreter Log:

 

Error: Reference "E" has no valid aspect.  Intersection of aspects the
referred models might have is empty.

 

I _know_ I've seen something like this before but I can't remember what the
problem was (is) and (or) how to resolve it.  I'm working on a smaller model
that duplicates the error, but in the meantime I'm wondering if anyone knows
off-hand what might cause this error to occur?  Has anyone seen this before?
I'm using GME 5.9.21.

 

Thanks for any insight that can be provided,

-ace

____________

Ace Thompson

Ignite Analytics Corporation

acethompson at igniteweb.net

m. 480 703 8909

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.isis.vanderbilt.edu/pipermail/gme-users/attachments/20060605/ef8e53b9/attachment.htm