enterprise distribution

[gaf at blu.org (Jerry Feldman)]

On Thursday 17 March 2005 8:20 am, David Kramer wrote:

> Yup.  That's one of the ways Agile/XP makes up for some of the extra
> process time: Catch more bugs early on, and it costs a whole lot yet. 
> Two of my partners were on an embedded project that used some of the
> Agile practices (Agile and XP did not formally exist at the time), and
> they had something like 15 bugs discovered after release in three years. 
> That's not bad, considering the complexity of the application.
>
> Also consider that finding and fixing bugs late in the development cycle
> is expensive, but finding and fixing bugs after the product is released,
> maybe even two or more releases later, *much* more expensive, especially
> when you consider the high turnover in a place like MSFT.
Maybe a bit OT, but...
I once worked for Higher Order Software (this was the company that coined 
CASE). The company was a spinoff of Draper Labs where the principals were 
doing a study of the software bugs in the NASA lunar project. They came up 
with a way to create "provably correct" software. James Martin wrote a book 
about the methodology. Essentially, they used the math function model:
y = f(x) where x is the inputs, y is the outputs. One of the nice things 
about the tools, was that a designer could use the tool to design and 
prototype. An implementer could take those "control maps" and build the 
underlying implementation. Some of the key software people from HOS founded 
Mathsoft. 

My thesis advisor made a statement about the methodology<
"The COBOL programmers hate it and are the ones who need it the most, and 
software engineers love it, but need it the least".

-- 
Jerry Feldman <gaf at blu.org>
Boston Linux and Unix user group
http://www.blu.org PGP key id:C5061EA9
PGP Key fingerprint:053C 73EC 3AC1 5C44 3E14 9245 FB00 3ED5 C506 1EA9