Managing graphical resources

Publisher Website
Google Scholar
Abstract
The development of interactive graphics software has separated into two apparently incompatible streams. The mainstream, exemplified by GKS and the Core, and a "RasterOp" stream, exemplified by Smalltalk and the Carnegie-Mellon CANVAS package. In the mainstream, the important concepts are viewing, segments, output primitives, and virtual input devices. In the RasterOp stream, the important concepts are windows, the refresh hierarchy, and pointing for input.Although it is easy in the mainstream to provide libraries of output-only routines for general use, it is difficult to do the same for input. Although it is possible to write device-independent interactive applications using mainstream techniques, the quality of the user interface may vary wildly as they are ported. These problems are caused by the mainstream's lack of the mechanisms needed to control the allocation of real graphical resources within the application:1. No hierarchy of pictures on the view surface. All parts of the application have control over all parts of the view surface.2. No connection between viewing and segmentation. The viewing parameters used to create a segment are not attributes of the segment.3. All parts of the application have access to all the input devices. It is up to each part to decide if an input is relevant, or if not, what to do with it.

This publication has 8 references indexed in Scilit: