CHIME: A Metadata-Based Distributed Software Development Environment

1
CHIME A Metadata-Based Distributed Software
Development Environment

• Stephen E. DossickDept. of Computer
  ScienceColumbia Universitysdossick_at_cs.columbia.e
  du

2
An Introduction

• Even small development efforts may involve
  hundreds of artifacts
• Design Docs, Change Requests, Test Plans, Code
  Reviews, etc.
• Sheer volume of data makes it difficult to find
  what you are looking for, hard for new project
  members to get started
• Other research areas may offer us some hope
• 3D Information Visualization
• Hypermedia

3
CHIMEColumbia Hypermedia IMmersion Environment

• XML-based framework allows software artifacts to
  reside in their original locations
• Source Code in SCM system
• Design Docs in a Document Management System
• Email archives on Corporate Intranet
• Users interact with project artifacts (and each
  other) by walking around a 3D Multi-User-Domain
  environment
• Novice users collaborate with experts who may
  be in different physical locations

4

5
Conceptual Model

• Groupspaces
• Persistent collaborative virtual space in which
  participants work
• Project artifacts as well as tools used to
  create, modify, and maintain them
• Artifacts continue to reside in their original
  repositories
• SCM systems, legacy databases, rationale capture
  systems, etc.

6
Conceptual Model (2)

• Groupviews
• Multiuser, scalable user interfaces used to
  navigate and work in a Groupspace
• Help users find and access relevant information
  quickly
• Keep users up to date on activities of other
  participants
• Virtual Environment rooms created from
  artifacts residing in the Groupspace
• Example Code module becomes a room, source files
  become furnishings in the room. Corridors might
  link a modules room to a room containing design
  docs, etc.
• Selective Awareness of other users actions

7
Conceptual Model (3)

• Software Immersion
• Virtual space defined by structure of artifacts
• Built semi-automatically from artifacts,
  relationships
• Architecture, organization of overall system can
  be inhabited, virtually toured

8
Conceptual Model (4)

• Challenges
• Artifacts added, modified, deleted, moved in
  underlying Groupspace
• How to reflect this accurately in Groupview?
• Remote repositories for Groupspace data
• May not support transactions, locking
• May be temporarily unavailable
• Complexity of underlying software project
• How to design a Software Immersion which users
  can understand?

9
CHIME Architecture

10
CHIME Architecture (2)

• Xanth Data Service
• Data organized into multi-rooted tree hierarchy
  of XML dataElements
• Each dataElement describes a single piece of data
  residing in a remote repository
• Xanth maintains an XML document which completely
  describes the contents of the Groupspace

11
CHIME Architecture (3)
ltdataElementnameREADMEid1000protocolhtt
pserverlibrary.psl.cs.columbia.eduport80
path/linux-2.0.36/READMEhiddenfalseparent
0behaviorGET /gt
12
CHIME Architecture (4)

• Xanth Protocol Plugins
• Implement retrieval protocols needed to
  communicate with data repositories
• May provide behaviors to perform actions on the
  data
• E.g. for HTTP, behaviors are GET, POST, PUT
• Xanth Link Service
• Provides typed, n-ary, bi-directional hypertext
  links among dataElements

13
CHIME Architecture (5)

• Virtual Environment Modeler (VEM)
• Parameterize each dataElement according to a set
  of Virtual Environment Types
• Base Types Component, Container, Connector
• VEM Types extensible for customization to
  underlying data
• VEM does not hardwire a user interface
• Provides only hints on how artifacts are related

14
CHIME Architecture (6)

• Theme Service
• Responsible for display, format, UI of Virtual
  Environment inhabited by users
• Theme Plugins
• Downloaded from server at runtime by clients
• Handle UI representation of Virtual Environment
  for user 3D, text based, etc.
• MUD Service
• Keep track of user locations, handle user
  authentication, communications between users, etc.

15
Where do we go from here?

• How can we easily populate and manage a
  Groupspace?
• Versioning and Transaction support for Virtual
  Environments
• Personal Software Process Experiment
• Methodology Immersion?

16
Related Work

• MUD/MOO Systems
• Software Development Environments
• Groupware systems
• Software Visualization/Algorithm Animation

17
Conclusions

• CHIME Architecture flexible, scalable
• Linux 2.0.36 kernel source experiment
• Methodology Immersion likely to be as useful as
  Software Immersion
• Upcoming User Studies at Columbia and USC