I think the final point of Richard is the variety of targets he is using routinely - and I quite understand him.
I am switching between 3 main target processor groups now, and I would be happy to use a similar-looking debug tool for all of them.

An another point, Michael, the lifespan of a product varies wildly. While one wouldn't expect support e.g. on a mobile phone's firmware for longer than 2 years (and I believe most often it is far less...) in certain industries it is far more than a couple of years. An extreme example: my father worked in a textile factory where they used weaving machines dated A.D.1890, still in perfect condition, producing quality fabric, requiring minimum maintenance - sometimes they had to replace a broken wheel or shaft, but that's simple machinery. If he would need some electronic control, he wouldn't buy it for that machine if he knew he would need to buy a new one within 10 years - the new one would bring no new functionality, as the machine itself could not be rebuilt for e.g. jacquard or jet waiving. Consider, they had tens of those machines, in a very small factory (a large plant has 100s-1000s).

As for myself, I am perhaps not too old (34) but I still don't like to change ANYTHING vital for my work more than twice a decade. If there is a successful and working design in an industry, there might be a lot of investment - and much more unseen than explicit - into the supporting infrastructure, than the potential benefit of a complete rework. Of course, you can always persuade the client that he needs the new stuff with all the bells and whistles, but if he is completely satisfied with the old one save a couple uf bugs and minor features... I think it is fair to provide support for it.

Personally, I barely wrote any substantial piece that lived actively less than for 5 years.

Jan Waclawek