Hi Michael,

Michael Ströder wrote:

HI!

when displaying an input form for renaming an entry I'd like to let the user search for possible new superior DNs. If applicable I'd like to construct the search filter derived from applicable DIT structure rules. So the question is how to find the structural object classes which are allowed for superior entries.

X.501 (1993) section 12.6.5 says: "Each object and alias entry is governed by a single DIT structure rule."

Currently I determine the governing structure rule by looking up the best matching name form for 1. the structural object class and 2. the current old RDN.

In general, you also need the governingStructureRule of the current superior entry, but it is easier to just look at the governingStructureRule of the entry to be moved. Ultimately though, the governingStructureRule of the entry to be moved doesn't matter.

Then I can derive the possible superior structure rules and lookup the superior structural object classes via the accompanying name forms.

But now I wonder what to do in case the user wants to change the RDN together with a new superior DN in one rename operation? From my understanding changing the RDN could lead to another governing structure rule. Is that right?

Yes, it could. In the general case the structural object class of the entry being renamed/moved would permit a number of candidate name forms, each with its own set of mandatory and optional naming attributes. For each name form, the structure rules specify the permitted governingStructureRule values for the new superior entry of the entry being renamed/moved. The structuralObjectClass of the superior entries doesn't come into it directly, but can be determined from the name form associated with the structure rule corresponding to each of the permitted values for the governingStructureRule of the new superior. It would be easier to search for entries having a governingStructureRule with one of the permitted values, and more precise since the relationship between structuralObjectClass and governingStructureRule is one to many.

The server will work out the new governingStructureRule for the renamed/moved entry from the governingStructureRule of the new superior and the naming attributes of the renamed/moved entry.

If you allow the user to select the new superior first, then that will constrain which name forms are available to rename the entry. If you allow the user to rename the entry first (explicitly or implicitly selecting one of the applicable name forms), then that will constrain which new superior entries are available.

Regards, Steven

Ciao, Michael.

P.S.: I'd appreciate interop testing over Internet regarding this...

Michael,

Michael Ströder wrote:

Steven,

I really appreciate your comment.

Steven Legg wrote:

...

Michael Ströder wrote:

...

when displaying an input form for renaming an entry I'd like to let the user search for possible new superior DNs. If applicable I'd like to construct the search filter derived from applicable DIT structure rules. So the question is how to find the structural object classes which are allowed for superior entries.

X.501 (1993) section 12.6.5 says: "Each object and alias entry is governed by a single DIT structure rule."

Currently I determine the governing structure rule by looking up the best matching name form for 1. the structural object class and 2. the current old RDN.

In general, you also need the governingStructureRule of the current superior entry,

You mean I could derive the possible structural object class from the governingStructureRule of the superior entry?

I was just making the point that an entry's governingStructureRule also depends on the governingStructureRule of its superior. However, you don't need to know the current governingStructureRule of the entry or its superior.

Hmm, there could be more possible structural object classes of superior entries.

...

but it is easier to just look at the governingStructureRule of the entry to be moved.

That's how I currently implemented it.

...

Ultimately though, the governingStructureRule of the entry to be moved doesn't matter.

...because the RDN could also change...

... because it only tells you about one valid possibility for a new superior. There can be others. Clearly, the entry could be moved under any new superior (in the same subschema administrative area) that has the same governingStructureRule as its old superior, provided the moving entry's naming attributes are not changed, because the structural object class, name form and superior entry governingStructureRule will all be the same in the new location. If the moving entry was schema-valid before the move it will be schema-valid after the move. By examining the name forms and structure rules in the schema it is possible to work out what other possibilities are valid, but they may necessitate a change to the naming attributes and may cause the governingStructureRule of the moved entry to change. The current governingStructureRule of the entry to be moved doesn't affect what's a valid possibility and what's not.

The possibilities in the schema boil down to a collection of pairs where the first item is an acceptable governingStructureRule value for a new superior and the second item is an acceptable name form for the entry to be moved, i.e., the first item tells you where the entry can go and the second item tells you whether the RDN needs to change.

...

...

Then I can derive the possible superior structure rules and lookup the superior structural object classes via the accompanying name forms.

But now I wonder what to do in case the user wants to change the RDN together with a new superior DN in one rename operation? From my understanding changing the RDN could lead to another governing structure rule. Is that right?

Yes, it could. In the general case the structural object class of the entry being renamed/moved would permit a number of candidate name forms, each with its own set of mandatory and optional naming attributes. For each name form, the structure rules specify the permitted governingStructureRule values for the new superior entry of the entry being renamed/moved. The structuralObjectClass of the superior entries doesn't come into it directly, but can be determined from the name form associated with the structure rule corresponding to each of the permitted values for the governingStructureRule of the new superior.

I could sort this out since I already have working code for determining the current name form of an entry and for determining the governingStructureRule. Not sure how to come up with a good user interface for that though.

Thinking about this a little bit more I came up with another difficulty: The new superior DN might reside in a part of the DIT where another subschema is defined. So this is a real blocker.

It means that the search filter is only valid within the scope of the subschema used to form it. It's rare for a single server to have more than one subschema area, so you could make that simplifying assumption, but in the more general case it means emitting one search for each subschema area the client knows about. It would help to ask for the subschemaSubentry to be returned so that matched entries that are not in the expected subschema area can be detected and ignored. Depending on the server implementation, you might be able to put a subschemaSubentry equality assertion in the search filter.

...

If you allow the user to select the new superior first, then that will constrain which name forms are available to rename the entry. If you allow the user to rename the entry first (explicitly or implicitly selecting one of the applicable name forms), then that will constrain which new superior entries are available.

Yes. But the problem is that it might be necessary for the user to change both in one rename operation.

I was assuming one operation, but allowing that it might be put together in a multi-step interaction at the user interface. Which step the user takes first constrains what can be done on the second step, but it isn't until after the second step that the "save" button gets clicked.

Presenting the user with a list of (superior DN, eligible naming attribute types) pairs to choose from would be one way of combining both steps in a single action. Of course, if the naming attribute types in the selection are different from the current naming attributes, the user interface should enter a screen to have the user edit them before allowing the operation to be submitted.

From my understanding changing the name form could lead to the situation where the old superior is not valid anymore.

Correct.

This all gets very hairy. At the moment I can't imagine a user interface which really helps the user to do the right thing without the user knowing a lot about how DIT structure rules and name forms work.

The governing structure rules are determined by structural object classes and name forms. A name form is basically a set of naming attributes, so the object class and "what attributes are used for naming" are all that the user really needs to be exposed to.

I will take a simpler approach with client-side configuration. Probably I will display a pre-configured select list of named LDAP URLs for searching the new superior entries. This would also work with servers which don't support DIT structure rules and name forms.

Whether or not you get back governingStructureRule is probably an accurate test of whether the server supports structure rules and name forms.

Regards, Steven

Ciao, Michael.