CMSCD3005

[ Home | Contents | Search | Post | Reply | Next | Previous | Up ]

Re: Methodologies

From: RMM
Email: @livjm.ac.uk
Remote Name: 195.92.194.14

Comments

>without ever needing to bother with

>A manager

Taking the example of the Linux kernel: Don't Linus Torvalds and Alan Cox have the final say as to what goes into the kernel ? If Linus says "no" - the answer is "no"... for now... ( he is likely to say "yes" if a good-enough argument can be made )

It isn't micro-managing of the kind Dilbert has to endure - but it *is* management in the sense that *someone* has set the agenda and general direction and then judges whether *other people's work* meets acceptable standards...

> A "development methodology"

Perhaps more accurate to say: "no single *mandatory* method OR methodology" - each hacker, group, company, whoever, is using the method OR methodology which which *they* feel most comfortable and/or productive

> A "life cycle" Sorry, I disagree - eg: updates to the kernel...

New feature required - support for USB hardware. New feature designed New feature implemented New feature tested New test version of kernel issued Iterate until reach pre-chosen mix of bugs fixed, facilities added, etc

New stable version of kernel issued to public

This *is* a life-cycle, but the necessary resources (ie: developers) are not all in-house

Question: Is Open Source a radical new concept or simply a more advanced form of outsourcing ? Justify your answer ( 30 marks )

> (without) Arcane diagrammatical techniques

Has any serious survey ever been conducted on this ? Attempt to correlate "quality" and "speed of production" of finished software with the method or methodology used.

Is "good" software produced by following a model method, or by a talented individual - as per Fred Brook's chief-surgeon - who adlibs a method for the job in hand ?

Departing into a serious question for a moment:

How much work is realistically involved in turning such speculation into a viable phD ?

RMM

Last changed: October 05, 2001