Requirements for AUS3.

General Goals

Referential integrity in data source

• Add an RDBMS - MySQL likely - that uses a schema designed to store the data required for business rules and update metadata -- including data points we may not even track currently (these need to be defined).

No duplication of data

• Basic normalization (without OVER normalizing) should quite easily take care of this.

Ability to automate releases (minimal human work)

• Dashboard that shows all valid releases and their updates would be beneficial for knowing the state of any given client that requests an update.
• Able to trigger new updates -- switch to say, "okay, push the udpate"
• Able to define fallbacks

Ability to automate tests and QA/verification

• Able to verify availability of update binaries, possible integration with Bouncer/mirror network
• Able to verify correct XML output based on incoming parameters
• Able to verify .mar application and update application

Automate .mar generation

• Able to generate .mar files by request? More from dmills here.