In general, I am not in favour of the top-to-bottom approach. Although in theory it seems to be more systematic approach than the bottom-to-top, but I have seen so many projects using the former approach suffering from wrong decisions (mainly underestimation of resources) which needed reiteration of the "top" work when working out the "bottom" details, that I rather simply omit that and start at the "bottom". It is kind of an "evolution" and I like things see to grow, rather than to happen. It is also encouraging for the developer(s), to have something working at the early stages of the project. Added bonus is, that if the client gets nervous, you already have something to show... :-)

In Mehdi's case, it's an answering machine - although a complicated one, with bells, whistles and kitchen sinks - so I would start with tying it up with the '51 and FLASH - forgetting keyboard, LCD and dunnowhat for now - just to record a message and play it back.

Of course, as everything, it is not black-and-white, so even if working bottom-to-top, some preliminary framework has to be done, identifying (well, in fact, guessing) bottlenecks and key issues, especially if more people are involved in the project.

Personal view, arguable.

Jan Waclawek