Ommitted components in the 1327 mapping

[Steve Kille <[e-mail]>]

I have been reviewing the discussion on this.   I'd like to review the
options, and I hope lead quickly towards a conclusion.  I can see
four choices:


1) Use a  attribute to indicate what happens at the lower levels.
Kevin suggested a filter mechanism.   This approach leads to a changed
algorithm.   All of the  other approaches retain the 1327 algorithm,
and give a mechanism to insert an (effective) null component into the
DN.

2) Use a special string.  Two variants:
  a) A short pathological string.  (e.g., " " has been discussed)
  b) Use a long name which is unlikely to clash with any real name
	(e.g, "*** Omitted Component ***")

3) Change the attribute syntax of all of the attributes to allow a
NULL variant.   


4) Define a new NULL component.   The value of the component is the
type that has been ommitted. This would lead to lookup of an address
of the form:
   C=US; A=XXX; NullComponent=PRMD; O=YYY



Let me suggest an analysis of this.   


1) Using an algorithm which does not follow 1327 seems to be a poor
idea.   It is also less efficient, and so I'd suggest not going down
this path.


2) This could be a fallback if nothing else was possible.  I think
that use of " " could be a problem, because it is also used as a real
value (at lease for ADMD).   I'm sure that we could agree a string if
needed.   However, I prefer to avoid approaches which rely on reserved
strings which we cannot enforce.


3) This is the closest to 1327 and the most elegant.   However, I
think that use of non-standard syntaxes for all the addressing
information is a very very bad idea.


4) I propose that we go with this approach.   It is simple and clean.



Steve