> Unless anyone can think of a good reason not to, I'd go with this. Seems
> logical enough and can provide good cross-referencing between modules.
I can think of several :-) (Well, I mean, he invited criticism :-)
A good design for the site should have the following qualities:
a) Obviousness. Given a document in your hand, it should be obvious
exactly where in the tree it goes in the vast majority of cases. I believe
that this design doesn't (yet :-) fulfil this criteria.
b) Balance. There shouldn't be root directories with two files in and
others with 2000, without a good reason.
c) Split along the fault lines. The divisions should be where they are in
people's heads.
> I'm sure if there are problems we can adjust this design to take them
> into account.
No, no, no :-) We finish the design as best we can _first_. This is very
important. Otherwise you get cruft. You'll be surprised how quickly design
decisions become very hard to change.
> Next problem is to identify exactly what information we have currently,
> and see if the above design works well with it.
No again :-) We write a design which a) can encompass all the current
information and b) provides room for expansion. We don't do a design and
then try and fit our information into it!
> While that is happening we should all raise our hands and be counted as
> site staff... Who other than Gerv is going to join me in that? All raise
> your left hand now! :)
Easy, there, tiger - I volunteered to do a small effort, medium reward
reorg. This is rapidly turning into a massive effort, big reward reorg.
I'm not completely sure I'll be able to spare the time. Not that it isn't
a great thing to do, and I'm glad I kick-started it, but I have final
exams next summer...
Gerv
