Re: case against XP (was: PG: Hackers and Painters)

["Neil W. Van Dyke" <address@hidden>]

Some impressions on XP, from talking with XP enthusiasts:

  1. Among other things, the XP umbrella codifies many pre-existing best
     practices (e.g., "unit tests are good").  Nothing wrong with that,
     though not all enthusiasts realize that XP didn't invent these things.

  2. XP adds to this particular practices that are especially appealing
     if you think that hard software engineering disciplines like
     requirements analysis, systems architecture, and project planning
     are *too* hard.

  3. Pair programming doesn't seem very applicable to situations in
     which a skilled programmer has to think hard about code,
     considering numerous issues and alternatives.  I imagine a pair
     programmer literally breathing down ones neck is counterproductive
     if one has to think hard.  I'd also assert that most code either
     needs a programmer to think hard about it, or is straightforward
     enough that pair programming is an inefficient use of expensive
     human resources.  I think there's more a place for efficient design
     meetings, and for sometimes designating people as on-call for rapid
     response to questions.

But how much truth is there to "Development:Languages::XP:LittleLanguages" ?

-- 
                                             http://www.neilvandyke.org/