An Assessment Framework for Automatic Code Generator Tools

摘要

This paper addresses the assessment Autocode Generators (AG) for their suitability for use in thedevelopment of software for critical systems. The assessment methods presented are independent of howthe AG was designed; therefore it is applicable to both white box (bespoke) and black box (COTS) Ags.The paper defines a number of attributes that relate to the quality of generated code from an AG and thecost effectiveness of using the AG. A number of metrics are derived to evaluate an AG against theattributes. For evaluation purposes, these metrics are applied to a number of COTS Ags.