Ok to me it boils down to this.
one developper tends to produce crap.
The reason is simple, code is not complex per se, what is complex is the
"big picture" holding all in your head so that you know where you are going
with stuff. Unfortunately the big picture, and it is a pity, is a big large
for a human brain... architecture is unfortunately a word that has lost its
original meaning.
I am not a fan of P-XP for "oh you forgot the ; here". I am fan of it in
the sense that it gives the BEST peer review early on, AND integrates the
views. It is the fastest way to get at the true nature of problems. Take
the CL discussion for example, it is a good example where the sum of our
inputs gives good result. Dr jung saw the problem but would have produced
crap by himself, what I proposed can be improved upon, Rickard I am not
sure what he brought, and what Dr Jung proposed today is closer but still i
am sure the kiddo will say something productive.
so that is the idea. Physical presence means nothing. It is the brain
presence, dematerialized, that is real. That is the real theory, much more
powerful than "boxes and circles" that mediocrities like to revel in. I
would argue that small groups is better for that kind of 'high-octane'
thinking, large groups tend to lose focus. Maybe that is the real theory
behind the discipline (and I mean the word) of Computer Science...
Right, so what I saying is that the "product" of our discipline is larger
than one brain (trivial to see in Operating Systems) and that many brains in
small tight groups are the way to do it. I remember the director of Xerox's
PARC talking about Open Source and him saying that we puppies didn't invent
anything (no shit shaft!) but that the one thing that intrigued him was our
capacity to crack the stability of very complex systems. OS's are there
today (see the problems MS has with Windows) J2EE is quickly getting there,
where BEA has to invest an immense amount of $$ to support it, long run it
is an adequate candidate for Open Source commoditizing.
It arrives at the non-trivial result that pair-programming and open source
in general enables some of the most "unmaintanable" infrastructure to be
maintained at lower cost than what proprietary vendors can do with the same
methodology.
marc
|-----Original Message-----
|From: [EMAIL PROTECTED]
|[mailto:[EMAIL PROTECTED]]On Behalf Of pohl
|Sent: Monday, January 15, 2001 12:07 PM
|To: jBoss Developer
|Subject: [jBoss-Dev] pair programming
|
|
|
|Since this subject was brought up, I wanted to ask how important
|to pair programming is the idea that two programmers be physically
|together in front of one terminal. Is it essential? If one
|can't do it that way, do you use instant messaging as a substitute?
|
|> Pair XP is the best discipline of Open Source Development.
|> Take myself and Rickard, Norbert/Hiram etc etc
|>
|> Reason is simple: we cannot follow everything. Pair is the way
|> to make sure you are not shooting yourself in the foot,
|>
|> In fact we should recommend that pairs be formed for all efforts
|> in jboss, that is the way to get the job done in open source
|>
|> etc etc...
|>
|> marc
|
|----
|[EMAIL PROTECTED]
|
|
