Mapping (was: Issues lists and the "preprocessing" topic)

John C Klensin klensin at jck.com
Mon Aug 25 02:24:16 CEST 2008



--On Friday, 22 August, 2008 18:06 -0700 Erik van der Poel
<erikv at google.com> wrote:

> This is just my personal preference, but I would prefer to
> have all normative material in protocol, table and bidi,
> preferably in an order in which you can read each document
> from beginning to end with minimal jumping ahead or jumping to
> other documents. The recommended reading order would be
> protocol, table, bidi and rationale, where the only "optional"
> document for an implementor would be the rationale. I would
> prefer to avoid duplication of text.

Once again, the problem is that there is a non-implementer
audience for these documents for which the [recommended] order
of reading ought to ideally be Rationale [, possibly Tables].

Obviously, one cannot have both your recommended order for
implementers and my recommended order for others _and_ have no
duplication of text, unless one adopts a model that no one has
proposed but which is probably feasible.   

In theory, one could introduce yet-another document.  Let's call
it 
"framework and definitions" for the moment.   It would contain
_only_ that normative material that was necessary to understand
both Rationale and any of { Protocol, Table, Bidi }.  The
reading order would then be

For implementers:
   Framework/Definitions, Protocol, Table, Bidi [, Rationale]
For others:
   Framework/Definitions, Rationale [, possibly Tables]

I would still prefer to defer trying to do this until we get
past Proposed Standard and go for Draft, simply because sorting
out what is normative and what is not is likely to be harder
than you (or Mark) think and I would prefer to not hold up the
WG while it is being discussed.  And "add a document" is clearly
counter to the beliefs of those who have been arguing that any
more than two documents (presumably Protocol and Tables) is too
many (and those who would like the reduce the number below two).
But this would be no harder to do than identifying the normative
material in Rationale and moving it to Protocol and it would
eliminate the "reading order" problem.

     john






More information about the Idna-update mailing list