On Wed, 24 Oct 2007 13:34:45 +1300 Brian E Carpenter <brian.e.carpenter@xxxxxxxxx> wrote: >On 2007-10-24 00:20, Simon Josefsson wrote: >> Norbert Bollow <nb@xxxxxxxxx> writes: >... >>> I would recommend that in order to be considered acceptable, >>> implementation in GPL'd free software as well as implementation in >>> proprietary closed-source software must both be allowed by the >>> licensing terms of any patents. >> >> I think that is a good recommendation, and I support it. >> >> I would even consider a requirement that in order to move beyond >> Proposed Standard, a protocol needs to have a free implementation >> available. > >There are two *very* different suggestions above. > >Norbert specifically suggests GPL compatibility as a requirement. >That is a very stringent requirement, because of the way the GPL >is written. Simon suggests the existence of a free implementation >as part of the IETF's implementation interop requirements; depending >on the definition of "free", that is a much milder requirement. > >In fact it seems like a quite natural extension of the rule >established in RFC 2026 section 4.1.2, first paragraph: >"If patented or otherwise controlled technology > is required for implementation, the separate implementations must > also have resulted from separate exercise of the licensing process." > >On 2007-10-24 08:06, Sam Hartman wrote: > >> Let me suggest starting with a lesser goal. Try to build a consensus >> that unless there is a good reason to do otherwise, it needs to be >> possible to write an open-source implementation of a standard and that >> the absence of such an implementation should be considered a red flag >> when advancing beyond proposed. > >s/red flag/yellow flag/ perhaps, but I agree this is a very reasonable >goal, and as far as I can see, essentially consistent with RFC 2026 >as quoted above. > >On 2007-10-24 02:58, Norbert Bollow wrote: >... >> That's IMO not quite strong enough. There are patent licenses which >> don't require to pay a fee but which impose other conditions that are >> so severe that having to pay a fee would be by far the lesser evil. >> >> How about: 'Should be possible to implement without having to ask for >> permission or pay a fee'? > >That will never fly. For good reason, many patent holders insist >on reciprocity conditions, and that seems to require an explicit >request and acknowledgement. And that will never fly (IANAL) with the GPL and so here we sit at an impasse again. So either a GPL implementation is important to interoperability in a given space or it is not. If it is important to interoperabilty, then this is a showstopper. If not, maybe not. >On 2007-10-24 07:57, Hallam-Baker, Phillip wrote: > >> If we get two RANDZ proposals and one thats only RAND we don't >> need to talk about the third one unless the IPR changes. > >I think we do, if the third one is clearly technically superior. >Why is the cost of a patent automatically more important than >*any* engineering cost or benefit? It's not the cost, but the incurred limit on interoperability. Scott K _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf