In normal SpecTcl you don't usually have to worry about data analysis works. You click buttons on a GUI or use menu items and poof SpecTcl is analyzing data. In normal SpecTcl, what's happening under the hood is a multi-step process:
An attach command stops any analysis in progress attaches SpecTcl to a new data source (file or program on the end of a pipe).
The ringformat command is used to indicate if the data source is NSCLDAQ-10.x or NSCLDAQ-11.x (we're not going to consider data from NSCLDAQ-8.x or earlier).
The start command is used to start analyzing data from the data source.
What this all does is select data sources, and buffer decoding objects as well as start a Tcl event loop based scheme for analyzing data while keeping the user interface alive to your interactions.
Batch SpecTcl does not need to keep the user interface alive and has been written with NSCLDAQ-11.x data in mind (though a suitably sophisticated programmer can override that).
Batch SpecTcl provides an analysis scheme were data are taken from some data getting object and passed to some data distributing object. In practice, the data getting object gets data from file while the data distributing object sends that data into the normal SpecTcl analysis objects.
To analyze data you must issue commands to select the data getter and the data distributor and then ask the program to analyze data. While this may seem overly complex, when we discuss parallel SpecTcl, we'll see how this allows us to extend batch SpecTcl into an MPI application simply by adding additional data getters and distributors as well as a fancier script to drive the analysis.
The filesource command specifies that the data getter the analysis will use gets data from a file. This command takes one mandatory parameter and an optional parameter. The mandatory parameter is the name of the file the getter takes data from. The optional one is the size in bytes of reads done from that file. When you run parallel SpecTcl, this block size can have an impact on when you go I/O limited. The default block size is 8192.
Here are some examples of the filesource command.
Example 2-3. Using the filesource command
filesource /mnt/evtdata/e00000/run6/run-0006-00.evt
filesource /mnt/evtdata/e00000/run6/run-0006-00.evt [expr 1024*1024]
In the first example, data will be gotten from the run 6 event file in the evtdata area for some experiment using the default block size of 8192. In the second example, The same file is read but with a block size of 1 Mbytes (1024*1024).
The analysis data distributor is selected using the analysissink command. It takes no parameters.
Once the getter and distributor are selected, you can begin analysis via the analyze command. This command will not return until the getter has indicated there's no more data available from its data source. Once that has happened, you're certainly free to analyze another run by using the filesource command to specify another file (for example you could analyze a segmented run in a loop over all the event files in the run).
Extending the filesource example above, here's how you would analyze the singly segmented run 6 with a blocksize of 1Mbytes