"James M. Polk" <jmpolk@xxxxxxxxx> writes: > At 04:48 PM 10/29/2007, Simon Josefsson wrote: >>"Eric Burger" <eburger@xxxxxxx> writes: >> >> > One interesting side effect of the existence of an open source >> > implementation of a protocol is monoculture. We ran into a problem in >> > ifax year ago when it turned out that all eight "independent" >> > implementations all relied on the same library, thus rendering the >> > "independent" implementations label difficult, to say the least. Why >> > were there no independent implementations? Because in this case, the >> > open source implementation was pretty good, and it was not worth >> > investing in a proprietary implementation. The result here has a really >> > bad side effect for the IETF: if there is a good open source, free >> > implementation, there will be no second implementation, resulting in it >> > being impossible for the standard to progress. >> >>But that is how it is supposed to work! If there is only one >>implementation, a standard is not mature enough to move to DS. You need >>to have at least two, preferably several more, completely independent >>implementations in order to quality-test a standard. > > but why does one or both have to be open source? > > Why can't both be commercial? DS designates a mature standard. If you read the requirements in RFC 2026 for a mature standard it is clear that few of the modern IETF protocols live up to that standard -- you need to demonstrate interoperability between two completely independent implementations of _all_ features in the protocol standard. Another (existing) requirement is that any patent licenses needs to be obtained through separate processes. I believe that a good way to demonstrate that the patent license process works is to require that a free software implementation exists. I strongly believe it should be possible to participate on the Internet without paying a software patent tax to some organizations. > So few PSs become DSs, I believe this will almost certainly make > progression from PS to DS slower. Is that what we want? I believe it will lead to better quality DS standards. The reason few PSs become DSs is, in my experience, not because a lack of free implementations, but rather that nobody cares enough to go through the pain of interop testing. The requirements for DS are pretty high already, which can be discussed as a separate issue, but this draft is only a marginal change. My impression is that your problem really is that few documents move to DS, not that a free implementation should be required. > I admit now all PSs have IPR attached, but this is almost certainly > intended to kill any IPR from achieving DS. Is that what is intended > here? I don't believe that was the intention, but that's a question for Brian. I disagree that all PSs are patented (if that is what you meant). I've implemented several such standards without worrying about patents. I believe the majority of PSs are actually published without known patents. A search in the IETF IPR tracker should answer that. /Simon _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf