AP-224 assemblies

[Lothar Klein]

Dave,

This topic is a very very ugly topic.
I only can recommend to you to study part43,
part41 (shape_representation_relationship) and also
what is written in AP214 clause 5.2.1.6 "Shape of an assembly".

My example:
There is a keyboard and a mouse on your table.
With shape_representation_relationship_with_transformation you
can describe the geometric relation of the keyboard and the
mouse. But the mouse is not an assembly of the keyboard and
vice versa.

As far as I can see the problem comes around with the compound
geometric models in AP203. This AP supports either ABREP or
manifold_surface or etc. - but is does not really support
a compound model consisting of e.g. a ABREP and manifold_surface.
And there is the RULE subtype_mandatory_shape_representation which
restricts this.
But the CAX-IF needed compound models and because of this rule they 
start to use the representation_relationship (in the early days
they did use mapped_item). 

With representation_relationship we
have the problem that it is NOT defined what is rep1 and rep2.
These attributes are equal. Several converters used rep1 for
the assembly, but meanwhile most use rep2 for the assembly.
This conflicts e.g. with the definitions of 
symbol_representation_relationship in part46.

Summary: 
- representation_relationship does not define an assembly 
- In AP214 there is no rule RULE subtype_mandatory_shape_representation
- The 214-ARM object TEMPLATE_INSTANCE is only mapped to mapped_item.
  We can't use representation_relationship for this.

Recommendations:
- don't change the current usage of AP203
- but change the CAD converters of AP214 when jumping to the
  IS version.
- continue to use mapped_item in AP224 
  (and this is also the case for AP210 and probably other APs as well)

Lothar

Dave Loffredo wrote:
> 
> While going through AP-224e2, I just noticed an bit of oddness about
> the way that it handles assemblies.
> 
> Right now, AP-224 mandates the use of mapped_items for assemblies
> rather than the item_defined transform available in AP-203 and AP-214
> and preferred (and extensively discussed) by the CAX-IF implementors.
> Can anyone verify my reading on this?
> 
> It should at the very least support both to be consistent with what
> the systems are currently putting out.
> 
>                                              - Dave

-- 
// Lothar Klein, LKSoftWare GmbH
// Steinweg 1, 36093 Kuenzell, Germany
// Tel: +49 661 933933-0, Fax: -2
// Email: lothar.klein at lksoft.com
// URL: http://www.lksoft.com