Re: short names, in general (Re: Comment on I-D ACTION:draft-ietf-ldapbis-dn-08.txt)

["Timothy Hahn" <hahnt@us.ibm.com>]

Kurt,

I can't tell what others "want", but I do know that "overlapping attribute
type names" (and object class names for that matter), is very real
"deployment" problem.

Every vendor/directory exploiter has come up with some set of "mitigating"
approaches - most of them based on Microsoft's Active Directory attribute
naming recommendations (i.e. prefix the attribute name with "company-", use
something specific to product within that, etc.)  But these are by no means
administered and/or followed and/or enforced.

Object identifiers are "nice" in that they allow for uniqueness and
de-centralized administration/management.  But they're particularly poor
with respect to user-friendliness.

Yes, I've heard all the arguments about "what flows on the wire is not what
people have to see".  But for every "transformation" that's required or
implied, it puts just a bit more weight on the shoulders of the, so-called,
"lightweight" client.  Unfortunately, "client" implementations that are
fully "schema aware" are few and far between - if they exist at all.  Add
to the problem the "feature" that depending on the area of the "tree"
you're "in", the schema to use may be different.  At this point, "client"
code throws in the towel.  More often, the "client" is merely a "conduit"
to get at what's flowing on the wire.

I guess I'd like to see either schema element naming recommendations,
guidelines, or a IETF best practice on the subject, along with room left
open for the so-called "extendable table" to be extended with new values as
appropriate.  If this means IANA should maintain such a table, then so be
it.

Regards,
Tim Hahn

Internet: hahnt@us.ibm.com
Internal: Timothy Hahn/Endicott/IBM@IBMUS
phone: 919.224.1565     tie-line: 8/687.1565
fax: 919.224.2540



                                                                                                                                          
                      "Kurt D. Zeilenga"                                                                                                  
                      <Kurt@OpenLDAP.org>         To:       David Chadwick <d.w.chadwick@salford.ac.uk>                                   
                      Sent by:                    cc:       ietf-ldapbis@OpenLDAP.org                                                     
                      owner-ietf-ldapbis@O        Subject:  short names, in general (Re: Comment on I-D                                   
                      penLDAP.org                  ACTION:draft-ietf-ldapbis-dn-08.txt)                                                   
                                                                                                                                          
                                                                                                                                          
                      08/28/2002 01:07 PM                                                                                                 
                                                                                                                                          
                                                                                                                                          



At 08:13 AM 2002-08-28, David Chadwick wrote (in part):
>I would like 5.3 to be edited as follows
>
>5.3. Use of Other Names
>
>  Attribute type names are not unique unless recorded in an
>internationally agreed document (such as an IETF standard like this one)
>or registry (such as IANA).

No.  RFC 2251 says:
   Attribute type textual names are non-unique, as two different
   specifications (neither in standards track RFCs) may choose the same
   name.

This means that there cannot be two "foo" on the Standard Track.
But there can be a "foo" on the Standard Track and multiple "foo"s
defined otherwise.

We'd actually have to mandate that only registered short names
(and private use names, e.g. x-name) be used in LDAP to eliminate
the possibility that a future registered name doesn't conflict
with preexisting names.

Is that what you want?  What do others think?

Kurt