> Isn't this just more evidence that it's very hard to specify interfaces in > unstable APIs?
Yep. And yet, why it's tempting to use them in certain ways. In full disclosure mode, these public API design questions interest me deeply for reasons of state not strictly related to Restlet, so I'll probably shut up and quit threadjacking in the interest of abstract knowledge. > But now you've got me thinking. What if (some day) there were a suite of > small interfaces like this: I've seen that pattern somewhere else recently, and for the life of me I can't remember where, but it was quite useful. In this particular case ... intuition tells me it could get noisy given the number of behaviors you'd have to provide interfaces for, but simply counting might prove that intuition wrong. In the end I'd just have to say +1 for descriptive, HTTP-coupled checked exceptions in the high level API ... for consideration in 2.0 *in conjunction with* decisions about where and what sort of interfaces should be exposed. I feel both of those changes are sufficiently dramatic and interrelated to constitute an interdependency. - R
