Ubuntu Wiki

• Created: September 20, 2005 by TomHoffman

• Priority:

• People: TomHoffman, NeedsSecond

• Contributors:
• Interested:

• Status: BrainDump, UbzBof

• Branch: n/a
• Malone bug: n/a
• Packages affected: n/a
• Depends: n/a
• Dependents:
  
  Display context of search results
  Case-sensitive searching
• BoF sessions: none yet

Summary

The question is whether or not Edubuntu ought to include the architecture necessary to support the Schools Interoperability Framework (SIF). Specifically, this would mean a Zone Integration Server and SIF "agents" for applications like Moodle, SchoolTool and OpenLDAP.

Rationale

The SIF standard is used in a growing number of school districts and states in the US, and may be adopted throughout the UK and elsewhere around the world. If Edubuntu shipped with SIF baked in, then key applications within a school's Edubuntu server (or servers) could share data, and Edubuntu applications could be integrated into the existing architecture of schools and vice versa.

Use cases

EdubuntuUserManagement -- A school sets up an Edubuntu server with OpenLDAP, Moodle and SchoolTool. They open up their Edubuntu SIF control panel and set SchoolTool as the authoritative source of the StudentSchoolEnrollment data object. Subsequently, whenever they enroll a student in SchoolTool, an account for that student is created in LDAP and Moodle.

Scope

Design

Implementation

Code

Data preservation and migration

Outstanding issues

One big question is how much developer time would it require to write a SIF-compliant Zone Integration Server (in Python using Twisted?) and several SIF agents?

Comment: note that we target only a single school in this release (which doesnt mean we shouldnt prepare SIF support) interoperation between schools will be a man target of dapper+1)

Response: While there are a few (attempted and planned) large inter-school deployments of SIF, in this case I'm only thinking of use within a single school. In particular, if we were going to push to put this in Dapper, our goal would probably have to be to produce components that match the spec as we read it and are simply tested with each other. There likely wouldn't be time to deal with broader testing and possibly formal SIF compliance. But it would be a big step to simply get an open source ZIS and some agents running out of the box on a Linux distribution.

BoF agenda and discussion