Re: [Idr] Update on advancement of rfc5575bis draft, implementations

[John Scudder <[e-mail]>]

Hi All,

In considering this further, I find myself sufficiently persuaded by Jeff’s position, in particular the shorter version he offered in a private note: "the level of rigor is perhaps excessive with regard to optional features”. Fair enough, in my opinion as a WG we should continue to be rigorous (ish) with respect to core aspects of a specification, but when it comes to optional features perhaps we can say non-implementation isn't blocking.

To be clear, I think in a perfect world a spec be fully implemented, and if it’s not, we pull out the non-implemented optional parts, put them in a different draft, advance the core spec, and leave the non-implemented parts in draft form until they become implemented. But that loads the authors and the WG with more work, so it may be better to respect the adage that “perfect is the enemy of good enough”. It’s true the WG has advanced plenty of other specs like this in the past (including RFC 4271!), for whatever that’s worth. The proposal is that as long as the non-implemented parts are genuinely optional and don’t impact the core feature or core BGP protocol, we’ll consider it potentially acceptable (though not preferable!) to proceed. 

If there’s any strong objection to advancing the document with this rationale, please follow up by the end of the week, otherwise I’ll go ahead and send the document to the IESG.

Thanks,

—John