Peter P. Jones wrote: (01)
>On 26 Feb 2004 at 5:41, Jack Park wrote:
>
>[...]
>
>
>>metal, so to speak. You know, when you are building a 7X7, that makes
>>sense. When you are building software, it makes almost no sense.
>>
>>
>[...]
>
>I wonder why that should be true. What prevents it making sense?
>Because as I see it, more _documented_ forethought has got to be
>good, hasn't it?
>
>
>
Maybe. Depends on the problem space. I suppose if some entity came to me
with a wishlist (e.g. use cases) for some enormous problem (here's one
from recent software disasters: the deadbeat dad project, where they
were sending checks to deadbeat dads instead of the needy mothers). When
someone gives you something like that, and when it is enormous, and
here, I'm just not gonna define "enormous", then maybe sitting down and
flowcharting stuff, appealing to all the design patterns you can muster,
and so forth, is a reasonable place to start. Actually, I do a bit of
that myself, even on small stuff. (02)
But, there is the mantra of XP, which is, among other things, release
early, release often. The nature of that is to get trial balloons out to
the users, with explicit understanding that they are, actually,
participating in the development process by telling you how the product
relates to their goals. Iterate, release often, etc. (03)
Maybe there's no real metric that says when to abandon a prior mental
machinations, but, in many smallish project cases, hacking design
documents absent any prototyping of major ideas, is probably just as
risky as just hacking prototypes absent design documents. Go figure...
Jack (04)
--
This message is archived at: (05)
http://collab.blueoxen.net/forums/cgi-bin/mesg.cgi?a=yak&i=403E57BA.5020505@thinkalong.com (06)
|