From: Samantha Atkins (firstname.lastname@example.org)
Date: Sat Apr 28 2001 - 16:08:16 MDT
Fabio Mascarenhas wrote:
> ----- Original Message -----
> From: "Patrick McCuller" <email@example.com>
> > A word of caution: extreme programming is not completely proven, and much
> > its advice goes against 30 years of software engineering practice. That's
> > to say it won't eventually be considered superior, only that there are
> > methods that are known to work reasonably well now.
> Well, XP is kind of like a complex system. The individual practices add up
> and reinforce it other. Take the "no upfront design" rule. At first sight it
> would quickly lead to programmers coding themselves into corners, but there
> comes Refactoring to correct the design as it's being generated, and there
> comes Unit Tests so the programmer can change the code at will and know if
> it still works and what broke, and there comes Pair Programming so another
> pair of eyes can spot some mistakes, and the list goes on and on.
> More and more projects are trying it, so more concrete results are already
Unit tests are written with at least a given partitioning of the problem
domain into language constructs, API and so on. If they are really
thorough they may also reach further into implementation details below
the encapsulation level. If implementation decisions and/or design and
even what is required are changing and evolving then the unit tests will
have to change also. This means that unit testing cannot replace design
or see if the system "still works" when the definition of "works" is
itself in flux.
Pair programming is a waste. It slows down productivity to the LCD of
the programmer's individual abilities minus the communication overhead.
It can work when a programmer/designer is stuck or needs a new view of
the problem but not as a general practice.
But I will refrain from further comments on this as it has nothing to do
with SL4 that I am aware of.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:36 MDT