In: Proceedings of the Fifth Workshops on Enabling Technologies: Infrastructure for Collaborative Enterprises (WETICE'96), Stanford, California, June 19-21, 1996. IEEE Computer Society Press. Pp232-237. Independent Validation of Specifications: A coordination headache Steve Easterbrook John Callahan steve@ atlantis.iw,nasa.gov [email protected] Software Research Lab NASA/WVU IV&V Facility Fairmont, WV 26554 Abstract Although we focus on IV&V, many of the coordination Large, complex projects face significant barriers to problems arefound in any large project. In many projects, coordination and communication due to continuous, rapid coordination problems surface merely as tensions and frustrations, perhaps leading to _ a_l budget changes during aproject's lifecycle. Such changes must be tracked, analyzed, and reconciled to ensure high-quali_ problems. In an IV&V process, the coordination crosses in the end-product, otherwise problems may. be get lost or ¢rganizational boundaries. This makes it harder forthe people involved to fred ways to work arcxmd anylackof ignored in the overall complexity. We report on "worLin- cooperation. Hence. theproblems are more visible, and progress" in the study of coordination problems between two independent, separate groups: software development easier to study. Within the IV&V process, we _tram on the and software analysis. We have begun to construct a taxonomy of coordination problem, which we illustrate analysisofrequirementsspecifications. This is a deliberate with two scenarios. We briefly describe current attempts to choice: independent analysis of requireanents specific- introduce incremental improvements to coordination ations has the potential for the biggest impaca of a.y problems in such projects via World-Wide-Web tools. IV&V activity. The quality of the analysis of design and Based on actual project experiences, we plan to deploy implementation and the rigor of testing depend to some such tools in a non-intrusive fashion to improve extent on the availability of good quality specifications. coordination and communication between software In this paper, we identifytlminterdew.ndenceoflarge volumes of documentation as a key problem in software development groups. specification. Any change to one part of a specifr.afion may havemany small impacts Cripples of influence" [6]) 1. Introduction throughout the documentation, which may be had to track down. Certainly this kind of problem was an important factor in the fatal decision to launda Omlleng_ [1]. Tim This paper describes some of the coordination mechanicalproblem that caused the blast was wellknown problems faced in the verification and validation (vav) before the accident; however, faiku'es in tracking the oflargesoftwaresystems.Inparticularw,e examine the problem and in keeping all docmnentation c,onsislent led processofIndependentV&V (orIV&V). a practiceused to faulty decision making by NASA managers. on large,safety-criticsaolftwaresystems inthedefense In this paper, we outline the 1V&V process and identify and aerospaceindustriesI.V&V facesmany ofthestone anumber of coordination problems that we have observed coordinationproblemsasany largedevelopmentproject, in actual IV&V processes. Specifically. we draw on our and introduces many problems of its own. In particular, informal observations of and imeractions with IV&V IV&V requires two organizations with conflicting goals to personnel working on the Space Station and the Earth cooperate to satisfy a single customer. Not only is there a Sciences Data and Information System (ESDIS) Im:_ets. lack of suitabletoolstosupport the ccmtxlinafion needed. We present two scenarios, which reveal the extent to but often even the infrastructure in which such tools whichcoordination problems can reduce the effectiveness would be used is missing. We will illustrate some of the of IV&V. We thea describe our initial work with the problems faced by IV&V through two scenaric_, and World Wide Web to provide infrastngtme and tools to describe our initial work on the development of web-based overcome some of _ problems. tools that address the problems.