Job Manager

With a simple analysis macro, you want to read in a file(s), do some processing, and then write out a file(s) (such a grouping of files, is refered to as a file set ). To read in and write out the files you have to tell the macro which ones you want (i.e you pass the macro the directory and run number of an input file and similarily for the output file). This works well for one user, using their own files on their own machine, running over just on set of inputs and outputs. But when you want to run on other machines (like the farms on RCF) or over multiple sets of input and output files, then you would have to change the macro to suit the environment you are running in + do other work to get it to run. Then you have many slightly different versions of essentially the same macro =>problems occur when tring to compare results etc.

The solution to this problem is the job manager object(TPhJobManager). It makes the macro an invarient, and puts all the changes into the jobmanager which is used as the input to an analysis macro. The job manager knows how to convert a machine independent reference name, that is used in the macro, to the specfic file name required on the place you are running. (e.g. ``RawData'' ->means ``/mnt/data/4/PhoRaw002306s000.dat'' on one machine but `` phobsink/engrun/data/PhoRaw002306s000.dat'' on another etc..). This way you do not have to change your analysis macro in any way when you change where you run (but you do change the job manager!). The other feature of the job manager is that it can contain multiple file sets that you can run over within your single analysis macro, ensuring that the macro itself does not change if you run over one file set or multiple file sets.

The details of how to use the job manager can be found at http://www.bnl.gov/phobos/Detectors/Computing/Documentation/jobmanager_doc.html/ , or goto Computing->Basic Docs->Job Manager Documentation (under Miscellaneous heading)