Drawing Guide
Last updated
Last updated
As you may have noticed, in many cases the best way of understanding a chain of processors is to represent it graphically.
A straightforward way of showing processor chains would be to depict processors as shapes, perhaps with a text label indicating what they do, and to use straight lines and arrows to illustrate their interlinking. Indeed, this is what is often done to represent composition in other systems, such as this example using Apache Storm Trident:
However, we found early on that BeepBeep processor chains illustrated in such a way are neither particularly intuitive (all processor chains look alike until you start reading what's in the boxes) nor very pleasing to the eye (after all, we have been producing color and graphics on computer monitors for at least thirty years). Therefore, we decided to develop a more colorful, intuitive, yet standardized way of drawing chains of processors.
In this appendix, the basic "rules" defining how to draw pipes, processors and functions are described; these conventions have been followed throughout this book, in the online documentation, as well as in all BeepBeep presentations given at scientific conferences in current years.
No need to be an artist to create your own processor chains. Using a vector drawing program such as Inkscape (freely available for all operating systems), it is easy to copy-paste the symbols from this book and include them in other drawings. A PDF document containing multiple pages of predefined symbols can also be obtained from the BeepBeep GitHub repository (look for a file called Drawing Guide).
The head of a pipe should indicate whether it is an input or an output pipe. This is done with the inward-pointing red triangle (for input) and the outward-pointing green triangle (for output).
In longer pipes, only the main body is longer; the head is not stretched.
The colour of a pipe should indicate the type of the events it contains. For the sake of consistency, we try to use the same colour for the same type across a diagram. At the very least, frequent event types should have the same colour across a common set of examples. Here are the colours that have been used for frequent event types in this book:
Pipe segments should be either vertical or horizontal. Orientation changes are done through rounded right-angle turns.
Pipe segments can be joined by vertically or horizontally centering them, overlapping them slightly, and using the Path/Union command in Inkscape.
Processors are represented by (square) boxes with input/output pipes around them. A symbol in the center of the box represents the processor's specific functionality. The colour of the input/output pipes should match the type of the corresponding input/output stream.
For a processor that takes parameters, these parameters should be placed across one of the unused sides of the processor's box.
For processors that have as parameter another processor chain or a function tree, a link to that object is drawn with a circle and a line.
The "cloud" can be replaced by a rectangle for better legibility.
Atomic functions are represented by rounded rectangles. When possible, their colour should have a similar shade to that of the input or output type of their arguments.
Function trees (i.e. composition of multiple atomic functions) are drawn as trees. For functions that need arguments from multiple input streams, the position of the stream is explicitly written in a lozenge. Stream numbers start at 1.
The edges can be collapsed if the resulting drawing is legible enough.
