[theforum] Can't get on steering, finance or sysadmin

[Martin Paul Burns]

Paul wrote on 23/05/2004 00:01:20:

> Martin Burns wrote:
> > Yes, it's called "never be too proud to admit the way that it used to
> > work best for you may not always be the case". And there are good
> > arguments for both more and fewer lists, so I expect it to be a dynamic

> > equilibrium for quite some time to come.
>
> Dynamic equilibrium = neverending cycle.

That's not necessarily A Bad Thing, but the way out of it is to stop
carping that there's the wrong number of lists :-)

> > Sorry, but as previously mentioned, the only sane reason you'd
> > custom-code a CMS from scratch now is if the CMS is an end in itself.
>
> Not necessarily so - I've worked on a number of sites (including mine)
> where we've built the CMS from scratch, including in some cases forums
> as well. It's difficult at first, but worth the effort in the end
> because of the flexibility you gain by not being tied down to a
> particular product.

Oh dear. How many CMS sites have you worked on with a packaged product?
Which packaged products? How many of the CMS you've written from scratch
have included i18n (the new site needs to have full i18n, including unicode
capability), version control, complex workflow, integration with any other
kind of system, role-based security with any kind of flexibility.
Additionally, how many of those self-build CMSs have been maintained and
developed once the original authors were no longer around?

The additional flexibility is largely an illusion. Most grown-up packaged
products have sufficiently flexible APIs that let you define all kinds of
new and different capabilities, content types, workflows etc.

Examples:
http://plone.org/documentation/archetypes/ - allows you to define new
content types as a simple UML schema
http://www.zope.org/Members/hathawsh/DCWorkflow_docs - allows you to define
states & transitions workflows without custom coding, and apply them to any
content type
I'm sure I could drag out a process-based workflow add-on too with a bit of
time.

Packaged products have also nearly always got better QA processes, more
developers whose work you can piggy-back,

The requirement for "Our site is unique and needs more flexibility than any
packaged product can offer" is usually a fallacy of vanity, and is easily
dismissed if you've done your requirements analysis with enough rigour.

btw, my day job was CMS consultant for several years, and these days I get
to manage system implementation projects

> I agree that using what we have or an off-the-shelf solution might be
> quicker in the short run, but I think in the long run it would be easier
> to have a custom-built system.

No, because then you end up with a system that only 3 people in the world
understand enough to do any work on. This is generally referred to as the
'under the bus' scenario.

I would also suggest that 'quicker' is a real benefit as we are volunteers
with limited time to spare. (Insert bitching about the 65 hour week I've
just worked, not including answering evolt mail)

> It's justifiable for something as big as
> evolt and with sufficient volunteers, in my opinion.

Ah, the "with infinite resources" fallacy. We don't have sufficient
volunteer time.

> I don't think we should be spending donated money on Windows licenses
> for servers

There have been some reasonable arguments put forward to suggest that the
cost difference isn't that great if you're paying for managed hosting such
as that offered by ServerMatrix, Rackspace, Rackshack etc.

Cheers
Martin