Introduction Echo is a computer language for scripting the processing of information; this can be in the form of structured data (databases, flat files, forms, spreadsheets, etc.) or unstructured content (documents, images, etc.). Processing includes actions such as transforming data from one format to another, accessing or storing content, correlating data, and generation of new content. Echo provides a set of commands for specifying these actions in such a way that the processing actions can be composed into information processing applications. Rather than attempt to redefine the functionality offered by XML processing languages such as XSLT and XQuery, Echo leverages these standards by defining a straightforward, implementation neutral, mechanism for expressing how the inputs and outputs of these processors should be tied together. Motivations The Echo language design was informed by these observations: • Much of what we do in information processing can be viewed as (or expressed as) transformations of data. ... Programs are filters that, given an input, produce an output. • Sophisticated, powerful, information processing systems can be built by composing smaller, special purpose transformation filters in processing chains. ... This is the pipes and filters model. • A simple, common data format (XML) that is capable of representing a wide range of rich, complex data structures is ideal for sending through the pipes between filters. • XSLT is excellent for defining many (but not all) transformation filters. • In order to compose filters into processing chains, a language is necessary to describe that composition. • By making that language, itself, representable in our common data format (XML), we can use processing chains to dynamically construct new processing chains, gaining additional expressive power while keeping the number of commands in the language small. • By connecting the filters with parse event streams, we gain in performance efficiencies without information loss. • An execution environment for the processing chains is needed to provide services such as authorization, connectivity and collection management. Echo was designed to provide application developers with a natural, direct way to express these pipelined processing chains. Echo contains commands for a rich set of filters and generators. Developers control the composition of the tasks and their order of execution simply by the commands' relative placement within an XML document, with leaf nodes sending their output toward containing elements (commands), which in turn may filter the data and send that output to their parents, upward towards the Echo document's root element and the Echo processor which invoked it. Echo allows for the implementation of extremely fast and efficient processors: • Processors may process large scripts as streams, producing output before having finished loading the entire script, enabling high volume, dynamic content applications . • As Echo is a functional language with limited opportunity for side effects, Echo processors can easily inspect scripts to identify results which may be cached for re-use, obtained in advance of their need, or bypassed because they will not be used in a particular case • Echo scripts may be compiled to byte-code or machine code
