Signature Matching: A Key to Reuse
Abstract
Software reuse sounds like a good idea. It promises advantages like reducing the time and cost spent on programming, increasing programmer's productivity, and increasing program reliability BP89, AM87, IEE84, Pre87. But why doesn't it work in practice? One reason is that it is hard to find things. As libraries of software components get larger, this problem will get worse. Reuse is only worth it if it is easier to locate (and approximately modify) a reuseable component than to write it from scratch. Today, if we want to find some desired component, we could use the component's name--if we are lucky enough to know, remember, or guess it. We could browse through the library itself, or perhaps an index into the library (for example, as with a Smalltalk browser). given that the components over which we are searching are program units (e.g., Pascal procedures, C Function, Ada packages, C++ or Smalltalk classes, or Modula-3 or ML molecules), then we have another means for retrieval: signature matching. This paper presents the foundations for what signature matching means and briefly describes a signature matching facility we have built and integrated into our local ML programming environment.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 03, 1993
- Accession Number
- ADA265332
Entities
People
- Amy M. Zaremski
- Jeannette Wing
Organizations
- Carnegie Mellon University