Re: people CN

[Andrew Findlay <[e-mail]>]

>Just to throw in my tuppence worth..., I should like to say that
>my preferred solution is to use a multi-value RDN.  For
>preference I would use organizationalUnit as the disambiguating
>attribute, although I accept that it has some of the same
>problems as already discusses, as I think that I have seen it

I would certainly support that. Any reasonable search operation will
then return a small number of sensibly-named entries for the user to
select from.

>recommended somewhere else.  There will always be exception
>cases when the disambiguating mecahnism does not use something
>like arbitrary serial numbers.

Right - for example we have two `Andrew Smith's in our Mechanical
Engineering Department. From our point of view the obvious
distinguishing attribute is year-of-entry since both are undergrads,
but it is probably not clever to use a locally-defined attribute in an
RDN. This leaves `Description' or `Title', which I think we can live
with.

It is almost inevitable that the two people with identical names will not
join the institution at the same moment, so one will already have a DN
assigned. I think it is *very important* that DNs are not changed once
they have been assigned except for very good reason. The arrival of a
name clash does not seem good enough, so I would expect to see the
first `Andrew Smith' keeping his DN, with the second one getting
something like `cn=Andrew Smith%desc=Undergraduate 1991 entry'.

All of this is easily solved with X.500 because it has search
operations and can return multiple values for consideration. A related
but more difficult problem is assignment of mail addresses. These
should also be stable and unique, but how should they be disambiguated
when both RFC822 and X.400 forms are expected to look `sensible'?

Andrew
-- 

----------------------------------------------------------------------------
|      From Andrew Findlay at Brunel University, Uxbridge, UB8 3PH, UK     |
| [e-mail]       [phone] or [phone] x2512 |
----------------------------------------------------------------------------