No abstract

Open-Source Sol~.-warc (OSS) development is regarded as a successful model of cnenuraging "natural product evolution". To understand how this "natural product evolution" happens, we have conducted a case study of four typical OSS projects. Unlike most previous studies on software evolution that focus on the evolution of the system per se, our study takes a broader perspective: It examines not only the evolution of OSS ystems, but also the cvolution of the associated OSS communities, as well as the relationship between the two types of evolution.Through the case study, we have found that while collaborative development within a community is the essential characteristic of OSS, different collaboration models exist, and that the difference in collaboration model results in different evolution patterns of OSS systems and communities. To treat such differences systematically, we propose to classify OSS into three types:Exploration-Oriented, Utilio,.Oriented , and Service-Oriented.Such a classification can provide guidance on the creation and maintenance of sustainable OS5 development and communities.

This paper describes our approach of using spatial hypertext as a means separated from an end representation for hypertext authoring. By taking advantage of the power of rich interpretation and constant grounding capabilities of a spatial hypertext representation, ART001, ART006, and ART014 use spatial hypertext as a means for authoring linear, hierarchical, and network structures, respectively. The role of the space of the tools includes controlling a structure and annotating a structure. The three prototyped tools have been developed to demonstrate what visual interaction design concerns need to be taken into account to integrate a spatial hypertext as a means with another structural representation as an end. The paper concludes with a discussion of what it means to separate representations as a means from those as an end in hypertext authoring.

Recognizing that programming is basically remixing, this chapter looks into the cognitive, social, and practical aspects of searching for and using existing code in a programming task. A code search mechanism undoubtedly plays an essential supporting role in a developers search for code in his or her own programming task. Supporting code search activities, however, demands more than code search mechanisms. At the same time, code search mechanisms also help a developer in a wider spectrum of programming activities. We present the anatomy of the cognitive activity in which a developer searches for existing code, and we propose efficacy and attitude as two dimensions depicting code search activity. We discuss areas of necessary technical and socio-technical support for code search activities in addition to code search mechanisms. We conclude the chapter by calling for a developercentered remixing-oriented development environment.

As interactive computer systems become popular, the study of pointing devices is becoming more important. This article discusses research on the efficiency of pointing methods which include the combination of a pointing device and its controlling software. One observation from this research is that the pointing efficiency is worse on a model with a smaller target size. Here, we propose a new pointing method—the Brake Type Method—in order to reduce the error rate. In this method, we try to make pointing easier through sufficient control of the movement of the mouse‐cursor. This method was evaluated in an experiment, which was also done with a digitizer (a popular pointing device for graphics). Conclusions indicate that the Brake Type Method proposed here is quite effective in reducing the error rate without losing other aspects of pointing efficiency.