Feature Naming Theory
List of: Discussion Topic
Subjects: Feature Naming
Contents: Kernel

A "feature" is some operation that creates or modifies some geometry, or is some geometry that results from an operation. It is this resulting geometry that needs to be named, usually in terms of the inputs. An example of a feature is a sweep. The inputs are a list of edges that make up a profile, and another list of edges that represents a path to sweep along. There are several outputs, most noticeably a face for each edge of the profile. The feature modeler is interested in which profile and path edges "generated" each face.

Feature modelers need to determine and maintain dependency graphs so that edited features can be regenerated along with any dependent features. Annotations facilitate maintaining the dependency graph to a high level of granularity. For example, a sweep operation may involve many segment profiles. However, editing one segment of the profile may be more efficient in certain circumstances than regenerating the whole sweep.

Primitive geometry creation, such as edges from a sketcher, are named arbitrarily e1, e2, e3, etc. Secondary geometry, such as what is created by a sweep, is named by reference to the primitive geometry. For example, Fs(e1, p1) would be the face created by sweeping the profile edge e1 along the path edge p1. Es(e1, p1) would be the edge bounding the far end of the sweep using the same inputs.

In a chained operation such as a sweep followed by a blend, the names generated in the early operation can be used in the later operation. For example, Fb(Es(e1, p1)) indicates the face resulting from blending the edge bounding the far end of the sweep of e1 along p1.

ACIS does not have any feature modeling software. It only has geometric modeling software, so applications must provide their own feature modeling engine.
PDF/KERN/08ANNO.PDF
HTM/DATA/KERN/KERN/08ANNO/0004.HTM