Ping-pong Pairing, it’s not just for breakfast anymore

Yesterday was my first day working at my new employer, Bayern Software, with (name drop alert!) Jeremy Miller.  We started right off with an easy story doing ping-pong pairing. I really dig this method of doing development (not to mention pairing, not to mention TDD).

We started off with a quick whiteboard discussion. Fortunately this story was so easy, there wasn’t much room for debate or design decisions, so we went right to the keyboard.  We were going to implement the Command Executor Pattern and start off with a synchronous implementation to make testing easier (and worry about the asynchronous stuff later).

Our setup this:  Jeremy at his laptop with a keyboard and mouse  and me at a monitor connected to his laptop with my own keyboard and mouse (also connected to the laptop).

How it worked

Jeremy started off and added an interface (ICommandExecutor) with some methods we know we’d need.  At this point I noticed he was using r-click, add new item in Visual Studio to add the new interface and I was able to suggest a ReSharper shortcut (hit ALT+INS and choose ‘Interface’) to have it generate a new interface file for you using a more refined template than the default one built into VS.  Score one for Pair Programming!  Sidebar for those playing the home game: that would be Chad 1, Jeremy 0 — the game would eventually end with Chad 1, Jeremy 5,984. He then added a new class called SynchronousCommandExecutor, had it implement ICommandExecutor, use R# to stub out all the methods (each throwing a ‘NotImplementedException’) to satisfy the ICommandExecutor requirements, and then he added it to the project’s StructureMap configuration code.

He then wrote a test for the first method we added to the ICommandExecutor interface.  It was very simple and straight to the point. He ran the test, which failed of course because all the method does now is throw a NotImplementedException. He then turned the control over to me and I started banging on the keyboard.  I went to the SynchronousCommandExecutor class and removed the ‘throw new NotImplementedException’, and then implemented the code of the method properly.  I then re-ran the test (using the Re-Run Last Test keyboard shortcut for R# which he happened to have bound to CTRL+3) and it passed!  Throughout this process, Jeremy kept correcting all my inefficiencies by showing me how R# can do that, or do this with a simple keyboard shortcut.  I thought I knew R# pretty well, but I didn’t. Shame! 

Note to self: We (.NET community in general) need more R# screencasts that show R#-fu in action.

Next, I wrote the test for the next method in SynchronousCommandExecutor.  I ran it, and it failed.  I then turned control of the system over to Jeremy who then implemented the method using a flurry of R# shortcuts.  He then ran the test again and now it passed.  

We proceeded in this manner until we had tested the SynchronousCommandExecutor thoroughly and were both satisfied that it was well tested. We then moved onto a different story that was more complicated and involved touching more parts of the system.

Why I like it

First, I got to tour the system in a non-threatening way.  I was able to discover the parts of the system I needed and work out from there. It wasn’t a heavy top-down or bottom-up scan through the code, it was focused on where I was concerned and branched out from there.  Second, Jeremy and I were able to swap (well, Jeremy was for the most part) tips and tricks to speed each other up.  There was a bit of friendly competition of sorts to see who could be most efficient.  It was mildly intense in a good way and broke the ice with a new codebase.  Third, perhaps most importantly, with in the first few hours on my first day, we had already completed a few stories. I learned more in those 2-3 hours than I would’ve learned after weeks of studying the code from the outside or pouring over reams of already-outdated “Systems Documentation” or specifications or requirements that are common at most non-Agile shops.  Fourth, by keeping each other honest and focused, we were less inclined to be distracted with other things or be tempted to floor the gas pedal and stream out a bunch of code that wasn’t tested properly.

Try it!

In most shops I’ve worked in (which were almost all non-Agile shops), pair programming happened. It may not have been institutionalized, but it was not uncommon to see two developers at one workstation hammering out an issue. It’s just natural, it works, so people instinctively gravitate toward that style of cooperation. Sure, you don’t do it every hour of the day (nor should you!). There is a time and place for pair programming.  I suggest that WHEN you find yourself in a pair situation, make the most of it and become super-productive by using the ping-pong technique (even if you’re not doing TDD, and you’re just banging out code you can still benefit from this!).

Related Articles:

Post Footer automatically generated by Add Post Footer Plugin for wordpress.

About Chad Myers

Chad Myers is the Director of Development for Dovetail Software, in Austin, TX, where he leads a premiere software team building complex enterprise software products. Chad is a .NET software developer specializing in enterprise software designs and architectures. He has over 12 years of software development experience and a proven track record of Agile, test-driven project leadership using both Microsoft and open source tools. He is a community leader who speaks at the Austin .NET User's Group, the ADNUG Code Camp, and participates in various development communities and open source projects.
This entry was posted in Misc, Programming. Bookmark the permalink. Follow any comments here with the RSS feed for this post.
  • erichexter

    So I was pairing with Kevin Hurwitz and the following became really obvious. That they are a million ways to use R#. I found that Kevin had about 25 R# live templates and when utilizing them can literally type about 1000 words a minute. I needed Resharper Jedi just to understand what he was actually typing. I agree that more R# screen casts are needed. That being said we came to the conclusion that he was totally disabled on my machine because he had mapped custom short-cuts. I showed him that I was in the habit of using the Menu short cuts i.e. Alt-R U N to run all tests ect. This is a very portable way to use shortcuts which allows me to pair with virtually anyone. Maybe this needs to be a separate post, but it did have some relevance.