Dex Tasks

Available Tasks

Dex task fall into 9 general categories listed and described below.

Each task is accompanied with documentation which describes it's usage. Additionally, you may extend Dex with your own custom task. We will talk about this in depth in a later chapter.

Task Documentation

Each task is documented in the following format:

• DESCRIPTION - A short description of the task's purpose.
• CONFIGURATION - A comprehensive description of what configuration the task requires.
• INPUT - Information pertaining to any input requirements that the task has with respect to the Dex Stream. For instance, a task may require that a dataset consisting of 2 or more numeric columns be passed into the task as input.
  • SAMPLE INPUT - And optional input sample.
• OUTPUT - Information pertaining to any outputs the task might have. Such as:
  • Changes to the Dex Pipeline Stream
  • SAMPLE OUTPUT - An optional sample output section.
• EXAMPLE USAGES - An optional section when one might use the component and how.

Task Pipeline / Workflow

Tasks are executed sequentially in what is known as the Workflow or Task Pipeline.

The initial task receives a state object as input, and returns a state object as output. The output from the previous task becoming the input to the current task. The initial task is passed an empty state object. Within our task documentation, we will simply refer to this as the INPUT and OUTPUT of the task.

This state object contains a current copy of the data within the pipeline. This data object is essentially a small table consisting of a header and a series of rows, much like a CSV or RDBMS table.

We will sometimes refer to this passing of data from component to component as the Dex Data Stream.

Workflow Execution

Example Workflow

Consider the following workflow:

1. Read CSV("sales.csv")
2. Searchable Table
3. Select Columns("MONTH", "SALES", "PROFIT")
4. BarChart()

Read CSV

The first task is an input task called "Read CSV". In this workflow, since it is the first task, it receives an empty input object. When executed, "Read CSV" will read and parse the "sales.csv" file storing the data in the state object and returning this state object as output. This will, in turn, become the input for the next task in the workflow.

Searchable Table

The second task, "Searchable Table" simply receives the output from "Read CSV" as input. This task does not modify the state object, but simply displays a searchable table on the input it receives and returns this state unmodified to the next component in the flow; Select Columns.

Notice that there are 5 columns in the data as when Searchable Table is executed.

Select Columns

Select columns selects the MONTH, SALES and PROFIT columns, in that order. It filters out the rest of the data, displays the selected subset of the data within the execution window and then passes the filtered data to the next component via it's output state.

The input to the next Bar Chart component will only contain 3 columns. "MONTH", "SALES" and "PROFIT".

Bar Chart

Bar Chart receives input with 3 columns and displays MONTH vs SALES and PROFIT and then returns the state unmodified back to the pipeline. Since this is the last task in the workflow, nobody cares.

Dex Data Stream

The dex-data stream is simply a consistent input/output structure of type DexData.

DexData can be thought of as an in-memory CSV file with some convenience methods thrown in.:

DexData

Constructors

Methods

Task Specification Overview

Each task is described with a short description, a screen shot where applicable followed by a list of options.

Input and output specifications are noted with a specialized shorthand described in the next section.

Input/Output Specification

Within the dex-data stream, the data within the table is stored as a string. However, certain task will expect data of stricter types. Here, we will descibe an efficient shorthand notation for expressing those requirements succinctly.