Re: Attribute Name Length Bounds

[Ryan Moats <rmoats@lemurnetworks.net>]

On Mon, Jun 16, 2003 at 11:55:51AM -0600, Chris Apple wrote:
| I definitely agree that we have both of those problems to deal with when
| considering what to do.
| Hence my original statement of not really knowing
| where it belongs either.
| 
| It is not my intent to open a "Pandora's box," but
| I am leaning towards wanting something in an LDAPBIS
| deliverable to cover this particular need because its
| a general problem.
| 
| Maybe the other folks who seemed to be expressing a
| similar leaning could speak up again now that we've had a bit more
| discussion? I'm referring to Rick, Kathy, and Larry. What do you think
| about where such
| a requirement belongs?
| 
| Chris.
| 
| >>> Mark C Smith <mcs@netscape.com> 06/16/03 7:12 AM >>>
| Chris Apple wrote:
| 
| > I maintain my view that this particular issue is of
| > sufficiently general nature to justify consideration
| > of adding a requirement to one or more standards track
| > documents. Some of those could be from this group. That
| > question I asked in a separate posting to the WG.
| 
| I will be honest and say that I am not sure where such a requirement 
| belongs. Lack of a minimum lower bound does make it difficult for anyone
| 
| to create and publish schema that may be used with all LDAP 
| implementations. That seems like a fairly general problem to me.
| 
| I also see Jim and Kurt's point about the problems that will arise if 
| the LDAPBis group descends down the path of specifying minimum lengths 
| for all of the LDAP protocol elements.
| 
| -Mark

Some technical data points:

1) There are *published, in use* schema that already have attribute and
objectclass names > 48 chars in length.  Therefore, a SHOULD/MUST upper
limit of 48 will break an existing implementation, and that's a no-no.
Further, using IANA considerations to drive such a limit is not appropriate
because of the whole 1.3.6.1.4.1 branch.

2) If I don't have a minimum bound to attribute names, then the only
way I can have interoperability is pairwise testing, which is horribly
inefficient.

Because of the point 1 above, I don't think LDAPBIS can take a pass without
noting the fact that *any* upper limit in the attribute/objectclass name size
runs afoul of the 1.3.6.1.4.1 branch.  That's my proposal for maximal size.

On point 2, if LDAPBIS takes a pass on minimal length, I view it as a
failure to promote interoperability.  I'd propose to see the minimal length
be 48 because of the IANA considerations.

Further, I think both of these are in fact protocol issues because LDAP
can and does use attribute names on the wire.

Ryan