>>But presumably there would only be one
>{c=US, sP=MA, l=Cambridge, cN="Steve Kent"}+rfc822="kent@bbn.com"
>True. If the intent were to create multiple leaf entries below the
single entry of this form then the justification would be better,
but I don't that's what people have been suggesting.
It's not that I particularly like compound attributes, but that's what
I was suggesting, rather than having the rfc822 be a leaf under the user name
DN. That's what I mean to imply by the plus sign -- exactly like most
people have been suggesting for name+serial. At least the rfc822 name
component has a great semantic content than a serial number.
Yes. But, what people were suggesting was a DN with ONLY a given name
and surname, plus an e-mail address. That's a bit odd from a
directory structure perspective. Obviously we would not use the first
two components as RDNs under a root, since there would be no good way
to separate the database base for people on a large geographic basis
using those attributes. If we used the e-mail address as the first
RDN it would allow for meaningful partitioning, but because the
address is a composite value, it seems wrong too. It would be more
appropriate to use the DC attribute to mimic the DNS structure, then
put the user name as the leaf RDN. In that case, though, the given
name and surname are redundant, right?