CODEWITHSUNDEEP
idedataflow
sheldon.backstack
sheldon.backstack

Reputation: 70

How do visual programming languages work?

I'm exploring the possibility of presenting a visual interface to users of an application that would allow them to enter some custom code/functionality in a dataflow-ish style (a la Yahoo Pipes).

I'm wondering how, in Pipes for example, their visual editor could work. Could the visual code be compiled into a textual language which to be stored in a database? Or could the individual blocks, connectors, variables, etc. all be stored in a database?

What about visual programming language IDEs like Microsoft's Visual Studio? Is the code interpreted straight from the visual interface?

Upvotes: 1

Views: 1059

Answers (1)

ern0
ern0

Reputation: 3172

What you see on the screen is the top of the iceberg. Components are different size, difficult or simple programs, with public interfaces. In dataflow programming, these interfaces are producers and consumers (outputs and inputs), so the component can visualised as a black box with pins on its input and output sides. When you connect pins (ports), you lead one program's output to another program's input. The components are pre-compiled for you, they are ready to run, you have just set their consumers (inputs) and producers (outputs) by connecting them. That's why they are black boxes: they're programs, which you can't change (except if you got the source code).

The components are designed to be connected to others. Some cases, components can run stand-alone, but usually they have to be connected to do the complete work. Basicly, there are three kind of components: - source: generates output (which requires further processing or displaying), - process: receives input, processes it, then passes it to further processing or displaying, - sink: receives input, displays or saves it, and doesn't pass it to anyone.

A typical complete dataflow construction contains a source-process-process-sink chain, where the number of process-type components can be even zero (the data generated by source is displayed by a sink component). You can think for that three components as they were one program before, but they've had broken, and you can now re-assemble them.

One of the most well known dataflow system is Unix shell. The CLI commands are the components. They're precompiled, you just define a chain by putting "|" between them. Also, most "source" commands can be used stand-alone, like ls, and most "sink" components can receive input from file defined as argument, e.g. more.

Upvotes: 2

Related Questions