[thelist] Specifications

[rudy]

> Ummm, if your friends don't use a spec of any kind,
> how do they know when they've fulfilled the clients needs?

exactly, that's the "got to be done" logic

> Not to mention budgeting, timeframes, feature creep...

exactly, that's the "waste of time" argument


let me speak from the developer's point of view

(i'm sure martin will be along shortly to refer you to a
couple of articles from the project manager's point of view)

in a dilbertian sense, developing specs for something that
you don't end up building is dumb

so is creating specs that do not guide or help the developer

but useful specs can only be produced by someone familiar with
development issues

and useful specs require more than the vague generalities you
see in ppt presentations (e.g. "pages will be created that link to a
database of product images and information")

sadly, the developers are often the ones called upon to
produce the specs, and this interferes with the time they
have available for developing

it is a myth that you can stop the world until the specs are
complete

it is also naïve to expect client requirements to stay as frozen
as the specs were a few months ago when the project started

the best approach?

loose specs and involved clients, and lots of interaction with them

oh, and project managers that will back you up and translate for
you when you need to tell the client "this is not possible, and anyway
it's stupid"

project management, in my experience, is often "get as much work
outta them as possible" versus "estimate the hell out of it so i don't
end up working nights too"

what would be nice is software that allows you to design an application
and the software will spit out the finished code -- in this type of
scenario,
the specification and development efforts are one and the same

but you have to be a big rich corp to have software like that


rudy