> I sue that person and win. Does this sound correct to you? Sure. That's plausibly reasonable. The API definition is significantly more detailed that just the

> I sue that person and win. Does this sound correct to you? Sure. That's plausibly reasonable. The API definition is significantly more detailed that just the method names. The metal box maker could easily have created their own API. But instead they use an API created by someone else. * Compiling against headers is an important issue. That would be bad. I raised this question in the last thread but it didn't seem to be an issue. * A closed S3 API seems fine to me. Amazon doesn't have an obligation to make switching easy. At the same time they'd have incentives to make their API open to encourage people to start using S3. Permissive open source software is exceptionally popular these days. If Amazon wants to write a closed API that presents an opportunity for someone else to write an open API. * I think we'd be in a much better place if software could NOT be patented but APIs were copyrighted. Implementation code is already under copyright. * I don't think "got there first" is a real concern. TCP and UDP both define connect(), send(), and receive(). But they are radically different APIs with radically different specifications. Thank you for your thoughtful reply. This is a good discussion.