Chapter 1. Introduction

This manual describes the batchfeature of SpecTcl that was introduced in version 5.2. Batch SpecTcl supports fully automated offline processing of event files. When coupled with the external mpispectcl and mipspectcl packages, this offline processing can be performed in massively parallel clusters enabling performance that is only limmited I/O throughput.

This document is organized as follows:

Let's get a few concepts straight before we dive in. Batch SpecTcl is a purely non-interactive version of SpecTcl. Unlike the "normal" SpecTcl (we'll call that just SpecTcl from now on), it runs on a Tcl interpreter without the Tk Graphical user interface package installed. Furthermore, once you begin analyzing data, the interpreter will be blocked from accepting future commands until that analysis is finished.

As with SpecTcl, to use batch SpecTcl you'll need to define and register an analysis pipeline. The same event processors you used in SpecTcl can be usedin batch SpecTcl, although they have to be setup differently.

Since batch SpecTcl is not interactive, the assumption is that at some point you've figured out the set of parameters, spectra, gates and gate applications you need and saved them to some configuration file. Prior to beginning analysis, you'll pull those definition into batch spectcl.

Finally, since batch SpecTcl is non-interactive, you can't see the spectra it creates. Therefore, after analyzing data, you need to write the spectra to file where they can be read into an interactive SpecTcl.

In the next chapter we'll start to dive into how to get batch SpecTcl working for your code. Note that since batch SpecTcl is purely serial, it's a good testbench for processing that will be used by MPI SpecTcl.