Meeting

[abez]

Outright I have an anti-XP bias. I have used it before, even done the pair
programming but I feel it's not appropriate for many tasks, especially
programs of 30KLOC or more.

C3, the Program that Kent Beck and Cronies worked on at Chrysler, failed after
4 years of development. This was the program which was used as an example in
the XP books. Chrysler refuses to use XP for any future projects.

I'm not saying all of XP is bad but when it's based on Refactoring, which
already classified as an antipattern one has to be a little skeptical.

Test first is great if you don't know what your interfaces look like. It helps
designing interfaces on a code level.

Tests are usually always good. Although Djisktra has said something to the
effect that testing shows the presence of the bugs, not their absence.

A contrived case

#returns half a number
sub half {
	return int($_[0]/2);
}

and your test is half(4) you get 2, but if you put in 5 you get 2.5. Obviously
you're supposed to test with greater coverage but this was a simple example.

Of course Donald Knuth has said, "Beware of bugs in the above code; I have
only proved it correct, not tried it."

Have a good day everyone :)
	abram

On Thu, 12 Dec 2002, Peter Scott wrote:

> Okay, Monday the 16th it is.  I'll send out another notice closer to
> the time.  I'll talk about using Test:: modules to follow the eXtreme
> Programming (XP) dogma^Wmantra^Wprinciple of writing tests before code.
> --
> Peter Scott
> Pacific Systems Design Technologies
> http://www.perldebugged.com/
>

-- 
abez----- ----- ------ - ------ -- ------------
http://www.abez.ca/ Abram Hindle (abez at abez.ca)
--- --- ------ --------- - - ------ --------abez