But this doesn't do that for IETF LC at all! Everyone
not involved in the WG gets just the same notice as now.
This is true.
What I hope is different is that drafts taking this optional
approach are higher quality, being based on running code.
This is a stretch, and that *unprovable* assumption also bothers me.
We have lots of running code that implements specs badly. We have
lots of running code that implements exactly what's in the spec and
misses everything that's missing (consider Martin's comment about
security and i18n). We have lots of running code that doesn't
interoperate with other running code. Your proposal specifically does
NOT require any testing of the running code, nor any interoperability
demonstrations (nor would I want it to).
Running code, when it's an organic part of the document development,
is undoubtedly a good thing -- it doesn't make everything right, but,
yes, it does do *some* spec validation and probably does help spec
quality.
But code that's written as part of a rote process, just to achieve
another check-box on the shepherd writeup and justify special handling
is not likely to provide any of those benefits.
Barry