From: Michael Richardson <mcr(_at_)research(_dot_)solidum(_dot_)com>
I would like to see the IETF continue to consider the ASCII
text to be the master.
I would additionally like to see the secretariat accept drafts
in some TBD XML markup as well as a corresponding ASCII. The provided
ASCII should match that which the secretariat produces, likely by
providing the XML->ASCII formatter on a web site, and basing it open
some open source implementation.
Authors should provide the ASCII because they should have proofread
that version as well.
The dual Postscript and ASCII RFC's show that such a plan is likely to
cause more harm than good. That dual track compromise sounded good at
the time, but turned out poorly. There are other precedents showing
dual languages don't work regardless of good intentions, from the DOD
equivalents of the early RFC's to the ITU's labors with natural language
translations. There is also the truth that every experienced programmer
knows, that two definitions of one thing absolutely never agree.
If the ASCII version of an RFC is complete, then the XML alternate is
not needed. If the alternate is needed even only for "illustration",
then the ASCII version is incomplete and so wrong. As others have
said "incomplete" as seen by people who neither write RFC's nor
implement protocols is completely irrelevant and inadmissible.
As someone else pointed out, the crux of this recurring thread
is the conflict between those who care about the form of RFC's
including whether they use "modern technology" and those who care
about the contents and function of RFC's as definitions of protocols.
Absolutely all of us have ideas for improving of the form of RFC's, but
it's a lot harder to say something useful about RFC contents. Thus,
those who have not yet found an opportunity to Contribute To The Standards
Process see opportunities in "implementing modern text preparation in
the IETF" (with "implement" as in "implement TCP/IP in a corporate
network"). It seems clear that some of those who are adamant about
replacing boring old ASCII are more readers than authors or editors.
Worse, I suspect a few have not spent much time reading the ASCII stuff
and would read little in any format, not matter how modern.
I've noticed the resounding silence of some who have made substantive
contributions that might facilitate a transition to using XML for ID's.
Why do is that? Could it be that they distinguish tools for writing
(e.g. XML authoring tools) from the form and utility of an RFC?
Vernon Schryver vjs(_at_)rhyolite(_dot_)com