RE: Réf. : RE: New RPG programmers

["Joe Pluta" <joepluta@xxxxxxxxxxxxxxxxx>]

> From: antoine.contal@xxxxxxx
> 
> So you mean you're criticizing Extreme Programming just because it has
its
> zealots?

No, I'm criticizing zealots who try to use XP techniques where they
don't belong, such as in large ERP projects.  The XP folks themselves
insist that XP works best on small, isolated projects.  That's because
release dates are "flexible", you make lots of interim releases, and you
need constant access to your customers.  In the real world, especially
among ISVs, that model simply doesn't fit.

On the other hand, it might fit great in a small company where you have
"nearly no annoying customers from outside the company" and you can
"release when something good is ready for releasing".


> It's a real pity, because teams doing XP with RPG could be a good way
to
> show how modern this language can be.

Very little of XP is new to XP.  The part that is most at odds with
current practices is the constant revising of the estimate based on new
knowledge.  We call that "prototyping", and it's typically done prior to
actually starting software development.  If you're still prototyping
after you've established deadlines, then you established your deadlines
way too soon.  Maybe that's what happens to Microsoft, I don't know.

Automated testing, on the other hand, is a nice feature.  I'd love to be
able to institute that on a regular basis.  The main probably with the
use in some of the more radical XP approaches is that unit testing
replaces detailed design.  Where we normally have two or three design
decomposition phases (functional to detailed to technical) with unit
tests and system tests, XP usually says "code to the functional spec".
Unfortunately, some people take this to mean "don't worry about the
design -- as long as the code executes correctly, you don't care what it
looks like".

Anyway, I'm not anti-XP.  It has its place.  Just not EVERYPLACE.

Joe