jwcolby
jwcolby at colbyconsulting.com
Thu Apr 19 09:25:03 CDT 2012
I am building an error tracking system for my Address validation system, the mongo system that validates the hundreds of millions of addresses every month. The system is so complex that I need to track all of the errors that can be generated for analysis and program control. So I am building an error class and error supervisor class. The system uses a clsManager which loads and sequences jobs, a clsSupervisor ( a job or table to be processed) and a clsProcess ( a chunk of a job / table being processed). The errors I am trying to track are errors specific to processing a supervisor or a process. So I have potentially two "levels" of errors, Supervisor and Process. In the end though they will all have a SupervisorID, and if the error occurred in a process they will have a ProcessID, after which they will have the same set of fields - a datetime, Program Error string, Exception Number (if any) and an Exception Message. I need to handle generating clsErr instances and filling out all of the properties, storing them to SQL Server, and loading any existing errors out of SQL Server as a supervisor or process loads. My question then is do I create a clsErrSuperv which handles the basic process but only has stuff for clsSupervisor generated errors, and then build a second clsErrProc which inherits clsErrSupervisor and adds on the additional fields for handling the process errors? Or just do a single class with the ProcessID field and load built in? I don't do any inheritance to this point and it sounds like this would be a good place to start. -- John W. Colby Colby Consulting Reality is what refuses to go away when you do not believe in it