Black Boxes Applied
- 2 minsIn a recent post I wrote about the concept of “Black Boxes” that I learned through “Structured Design”. Yesterday we got to applied it with my student apprentice and the effect it had on us was quite remarkable.
We started building a small program that would take input from the user and search a file which would be parsed for a result. We started talking about what would be a good point to start with our tests but we were both not sure where to begin. At that point I pointed to the white board. With a pen in hand I started asking my apprentice as to what functionality we’re aware of. “We need to parse the data” he said. I drew a box that had a label ParseData. On the top of the box, I drew an arrow pointing down with the label “raw_data” and underneath the box an arrow pointing downwards with the label “formatted_data”. This represented a black box with a known input (raw_data) and a known output (formatted_data). The goal of that module from a high level view was clear:
We repeated the same thing for other pieces we felt they were necessary (ConsoleIO, QueryData) following the same steps as above. By the end of it, the pieces of the solution were well defined.
The biggest takeaway from the experience however was the fact that we decided upon a different starting point than what we initially planned on. Our thought was to perhaps start with the parsing of the data. However after drawing the boxes and discussing their different properties, it became clear that the starting point should be the QueryData for two reasons. For one, it was actually simpler to tackle than dealing with anything that involved parsing. The second point though which was the most important was the fact that by starting on the QueryData we could describe in our terms and in our tests how we wanted the parser to behave. The last point helped us identify very clearly what the ParseData module should give us and thus we focused on getting that desired output.
This is a powerful technique which can help break down all kinds of problems regardless of size and complexity. Applying it consistently is the tricky part…
Personal blog of Makis Otman