Kanga++ Technical Overview

Kanga++ Technical Goals

• Permit direct access to the persistent data stored in a ROOT file.
• Allow users navigate easily between data stored locations. This could mean different branches, trees or files.
• Make it simple to add "User" information to the official store.
• Make it possible to split data by analysis stream or detector component.

Kanga++: Storing Data Objects and References.

• To permit direct access to the persistent data, we seperated out each of the unpacking functions used in reading the data, so that they can be called directly by the user.
• To allow users to navigate between data stored in different locations we defined a way of storing and "de-referencing" pointers persistently.
• We took advantage of the fact that Kanga++ data are stored on TTrees to give the user access to the TTree itself and let them add their own data.
• In order to be able to split the data by analysis stream or detector component, we insured that persistent references are valid across TTrees and even across files.