RootCnvSvc Code Review Part II

QUESTIONS

What is the time line for adding event selection based on trigger and run/event ids?

What is the schedule for adding converters for relations?

Is it acceptable to remove the RootSvc package and add that code to RootCnvSvc?

Do we know why the end-of-line characters are coming out incorrectly - what editor is being used and how can it be fixed?

Can the code currently in RootSvc be modified to use maps rather than vectors to access the TTrees?

Use exceptions rather than depending on the Gaudi log to report errors in RootSvc?

Do we want to create another package that will contain the details of conversion? This code could then be shared between RootIo and RootCnvSvc during the transition.

What recommendations does Ursula have for improving the structure of the TDS to make conversion easier? 

RECOMMENDATIONS

Complete testing on Windows and Linux

Fix requirements file to work on Windows

Fix end-of-line characters in the source

Add capability to handle selections for the trigger

Allow for random event access using run/event ids

Write converters for the relations

Write converters for LDF-specific data objects:  LdfTime and diagnostic data

Remove the RootSvc package and merge with RootCnvSvc

Modify scripts in RootCnvSvc to be more re-usable

It would be better to modify RootCnvSvc so that it does not call the McEventCnv, DigiEventCnv, and ReconEventCnv directly.

Bring the mainpage and release.notes up to date, specifically add in description of jobOptions in the mainpage, including the default values.

Add one or two UML diagrams about dynamic behavior (collaboration diagram and/or sequence diagram)

Add a reference to the Gaudi documentation concerning converters and the conversion service.

Extend the Doxygen documentation to further explain how ROOT access is performed.

 

For the Core Group to consider

Re-evaluate the TDS and ROOT event classes and structure.

 

Back to GLAST Software Home

H. Kelly Last Modified:  2004-08-04 15:58:56 -0700