Automated dependency resolution for open source software
Title | Automated dependency resolution for open source software |
Publication Type | Conference Paper |
Year of Publication | 2010 |
Authors | Ossher, J, Bajracharya, S, Lopes, C |
Secondary Title | 2010 7th IEEE Working Conference on Mining Software Repositories (MSR 2010)2010 7th IEEE Working Conference on Mining Software Repositories (MSR 2010) |
Pagination | 130 - 140 |
Publisher | IEEE |
Place Published | Cape Town, South Africa |
ISBN Number | 978-1-4244-6802-7 |
Keywords | dependencies, java, source code, sourcerer |
Abstract | Opportunities for software reuse are plentiful, thanks in large part to the widespread adoption of open source processes and the availability of search engines for locating relevant artifacts. One challenge presented by open source software reuse is simply getting a newly downloaded artifact to build/run in the first place. The artifact itself likely reuses other artifacts, and so depends on their being located to function properly. While merely tedious in the individual case, this can cause serious difficulties for those seeking to study open source software. It is simply not feasible to manually resolve dependencies for thousands of projects, and many forms of analysis require declarative completeness. In this paper we present a method for automatically resolving dependencies for open source software. It works by cross-referencing a project's missing type information with a repository of candidate artifacts. We have implemented this method on top of the Sourcerer, an infrastructure for the large-scale indexing and analysis of open source code. The performance of our resolution algorithm was evaluated in two parts. First, for a small number of popular open source projects, we manually examined the artifacts suggested by our system to determine if they were appropriate. Second, we applied the algorithm to the 13,241 projects in the Sourcerer managed repository to evaluate the rate of resolution success. The results demonstrate the feasibility of this approach, as the algorithm located all of the required artifacts needed by 3,904 additional projects, increasing the percentage of declaratively complete projects in Sourcerer from 39% to 69%. |
DOI | 10.1109/MSR.2010.5463346 |
Full Text |
- Log in or register to post comments
- Google Scholar
- DOI
- BibTeX
- Tagged
- EndNote XML