Re: [Fsfe-uk] QnetiQ study

[Chris Croughton]

On Sun, Aug 24, 2003 at 11:00:25PM +0100, Paul wrote:

> In other words, we've not really looked at the wealth of software out
> there, but decided to playing "pong" would be a good idea...

So this is new?

> What? This makes no sense. How can OSS be "closed shop"? The whole idea
> of OSS is that it is that - OSS.

You noticed that as well?  I wondered how on earth /open/ source can be
described as "closed shop".  Unless they are looking at developers seen
as cliqueish (which is true to an extent, but irrelevant because so are
all groups of people and OS developers probably less than most).

> > 3. There is no single consistent implementation control and
> > management methodology being operated by the case studies included
> > in this report. Each operated their own methodology, typically based
> > around the experience of those taking management control of the
> > implementation.
> 
> Not sure what they're on about here. Take 4 distros - SuSE, RedHat,
> Slackware and Mandrake. Each distro has XFree86 4.3, Gnome2 and a kernel
> around 2.4.20. They have the same libs and mostly the same core apps. 
> The kernel is implemented the same on all machines and usually, so is
> the directory structures. As to each operating they own methodology
> (assuming they're talking about people here), that is a good thing.

You missed "implementation control and management".  They are perfectly
correct here, there is indeed no consistent methodology being used in
either area, and for many OSS and FS projects there is little if any
'methodology' being used apart from "I think this is cool, I'll write
it" (see Mozilla, for example).  For some of the large ones, like the
Linux kernel and GCC, there is an element of implementation management
(things don't go into the released kernel unless Linus says so, for
instance) but still not much recognisable as as a 'methodology'.

Certainly some do use methodologies and tools like CASE and UML, but
it's certainly not consistent.

Now, whether such methodologies are worth the paper which describes them
is a different matter, but it is still something which a lot of
companies do look for.

(Can you imagine any OSS project passing ISO 9000/9001?)

> Windows does what it needs to (sometimes, on days ending in a 3, on odd
> weeks and if the moon is in alignment with both Jupiter and Mondas), but
> gives the user no choice. With Linux the manager can tailor each update
> to each machine, so while core apps get updated, if only one person is
> using OOo and another KWord, then the updates can be altered to reflect
> this.

It's nothing to do with "implementation control and management", though.

> To me, this makes for far better management and less bulk.

Again, irrelevant, what companies are looking for is the formal methods
they have come to expect.

> > 4. None of the case study organisations knew of a recognised
> > implementation methodology that they could use.
> 
> I have a feeling that they've gone to UK higher education of Microsoft
> for this!

See above.

> 9. Further studies should be conducted by people competetant to do it
> and who actually know what they're on about.
> 
> The document has the words "Microsoft" smothered all over it.

Er, no, it has "corporate procurement" through it.  80% or more of the
job adverts which arrive in my mailbox want CASE, UML, Yourdon and other
design and implementation methodologies.

Chris C