Loading ...
Sorry, an error occurred while loading the content.

[XP] Re: Re: [XP] Pair-programming -- You want to hear something real ly eXtreme!

Expand Messages
  • Randy Ynchausti
    Well, Mats recent questions about tri-programming/pair-programming have caused me to go back and briefly analyze some data to try and put some numbers to the
    Message 1 of 1 , Mar 15, 2000
    • 0 Attachment
      Well,

      Mats' recent questions about tri-programming/pair-programming have caused me
      to go back and briefly analyze some data to try and put some numbers to the
      pair-programming benefit. In trying to improve my personal software
      development process I do some things that seem pretty extreme to most
      developers.

      Anyway the data that I analyzed is from about 2 weeks worth of development
      where I did not work in a pair. During that period I wrote about 600 lines
      of Smalltalk code for an XML processing system with an accompanying 800
      lines of test code. Part of the process is to log all defects. A defect is
      defined as anything that does not work. That means anything that does not
      work. So I have a log of entries like:

      1) Misspelled a local variable name.
      2) Assigned a value to a temporary variable to be able to check that value
      -- but forgot to change the testing block of code to use the value.
      3) Created a temporary variable but never used it in the method.
      4) Sent a message to a class when I wanted to send it to a temporary
      variable
      in the method. This occurred because I copied the code and pasted it
      without reviewing it completely.
      . .
      . .
      . .
      n) It must be getting late. I implemented an at: without parentheses.


      There is a ton of data that I logged, but this gives you a picture of the
      detail in the comments. When I analyzed the comments to see what types of
      defects I created during that period -- I counted 11 out of 109 that were
      not things that I would expect a pair-programmer to have helped prevent.
      Now, since I spent about 8 hours fixing these 109 defects, that's about 10%
      more free time.

      You are right if you are thinking that this does amount to a hill of beans
      and it is not real concrete proof... I hope someone with more concrete data
      will respond. Otherwise, you will have to wait for me to do a little more
      testing (pardon the pun) and a lot more data analysis.

      Regards,

      Randy
    Your message has been successfully submitted and would be delivered to recipients shortly.