What ever happened to Tandem and NonStop OS 2006

Don Chibutton

Well, yes. I don't think we disagree.

Unfortunately, many managers are reluctant to let contracts that do not promise to deliver a defined product by a fixed date at a fixed cost, even if history shows that such targets are rarely achieved. They do not see why building software should be different from building bridges or roads, and that is a good point.

In the part you snipped, I wrote about the dynamic nature of requirements and the difficulty of foreseeing every detail. It has always been recognised that systems evolve - requirements change - after delivery, so it's reasonable to buttume they will change during development too, so design should be flexible enough to easily accomodate change. Iterative is certainly better, but should not dominate the process so as to make time and cost estimation too hard. OTOH, prototyping has it place in demonstrating what the sytem will look like and revealing mistakes and misunderstandings before undoing them becomes prohibitively expensive. I hadn't seen the revised version of MMM, and I'd taken "building one to throw away" as still valid in the prototyping context. Prototypes, built quickly and cheaply, are generally not suitable to be used as the structure for the final product, and trying to use them for this is asking for trouble.

--brian

-- "What's life? Life's easy. A quirk of matter. Nature's way of keeping meat fresh."