04-Nov-97 JEC

Scripts for FD Rope jobs

All scripts for FD rope jobs now have a standard structure, and standard argument list. A template to create a new script can be found in ~jec/jobutil/fd_job

Each job has an interactive phase, which creates the working directory and copies or links the required files into it, followed by a batch phase which compiles, links and runs the executable.

Interactive phase

The first five arguments of a standard script are

Period number
Pass number
Data type {cs,s,d}
FILE card type {Cat,Fat,Dat,Dad}
Number of events

The period, pass and data type can be found here These are passed to opaldsn for processing. If the inter script returns $INTER = T then the job is prepared for submission :-

A working directory is created under ~/temp7
A variable $Filename is defined using the job name, data period and pass number
The opaldsn script is executed to write the DATA, TRIGGER and CAL ffread cards to the file dsn.ffr, which is then copied to fort.4
Links to the opcal directory and the catalog are inserted
The PATCHY cradle is copied from the Job directory to the working directory
The FFREAD cards are copied from the script to the working directory
The job is submitted to the batch que defined by $Q ( = {wsvs/wss/wsm }) The output file is stored as ~/list/$Filename.lis, and histograms as ~/hist/$Filename.hrz

Execution phase

When ( if ? ) the job is eventually selected for execution on shift, the standard OPAL utilities are used to prepare the executable file :-

ytofort prepares the .f file using the cradle
egrep searches the Patchy listing for useful lines
fort compiles the .f file
link prepares the executable file. The link libraries are stored in the script
The executable is executed, using the FFREAD cards as input
The file 'rope.histo' is copied from the temporary directory to ~/hist/$Filename.hrz
The temporary files are all deleted

Running the debugger

If you ared so desparate as to want to run the debugger, the least difficult method is to comment out the lines at the end of the script and submit a short job. When it finishes, run the executable file in the temporary directory.

Variables defined in the standard script

Jobdir

The directory in which the script is stored

Job

The name of the script

Tempdir

The temporary directory in which the job is run

Period

of the data to be processed

Pass

of the data to be processed

Filename

for storing log and histogram files; setenv Filename "$Job"_"$Period"_"$Pass" WARNING - do not forget to change the definition of Job when editing a new script from an old one. If you forget, the new script will cause the old script to be submitted for execution.

Utilities used by the standard scripts

These utility scripts are stored in ~jec/jobutil :-

opaldsn: constructs the FFREAD cards required to access data on shift
inter: determines whether a script is running interactively or in batch
opalexpt: determines the Experiment number, given the period number