[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Non-existent syntax referenced (ITS#2029)

At 08:52 AM 2002-08-18, Michael Ströder wrote:
>Kurt D. Zeilenga wrote:
>>Applications should use subschema discovery to check whether
>>the server supports the schema elements they intend to use
>>and, hence, already knows.
>Well, let's assume an application has knowledge about how to deal with attribute type It wants to know whether 'teletexTerminalIdentifier' is supported by the server.

It is partially supported.  The server can return the attribute,
but cannot perform any action requires knowledge of the particulars
of the syntax.

>It happily reads the declaration from attributeTypes in sub schema sub entry. But the syntax used in the attribute type declaration is not supported by the server.

The server is not capable of validating values of this syntax.

>The conclusion by the application MUST be:
>=> The attribute type is not supported by this server because the client application has to assume that the server does not even know how to handle data with syntax (not to speak of matching).

Wrong conclusion.  The client should assume the server won't
accept values of this syntax.  It shouldn't assume the server
wouldn't provide values conforming to this syntax.

>=> All object classes containing the attribute 'teletexTerminalIdentifier' are not supported.

No. You can certainly add entry whose object classes allow
(but don't require) this attributes as long as this attribute
isn't provided.

>If you are referring to non-existing syntaxes you should leave out attribute ldapSyntaxes in sub schema sub entry completely. In this case an application could assume that the server does not make any statements about which syntaxes it supports and reasonable defaults could be applied.

Leaving out the attribute type would imply that the server
would never return this attribute type.  The server can return
this attribute.

>>  Applications should not use
>>schema which they don't have any knowledge of, except for
>>some definition provided in a subschema.
>Can you elaborate on this? Phrases starting with "except for" usually opens a big can of worms...

Applications should not use schema elements unless they
know the semantics of those elements.  Schema definitions
(provided in the subschema) do not convey semantics.