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

Re: Models: updating DIT Content Rules



At 01:57 AM 7/23/2003, Jim Sermersheim wrote:
>If a structural object class exists without a DIT content rule, and entries of that structural object class exist in the DIT, what are the rules for later adding a DIT content rule for that structural object class? The same issue exists for updates to existing DIT content rules.
> 
>Obviously, a DIT content rule which would invalidate existing entries could/should not be added. This could happen due to the addition of MUST or NOT attributes or removing AUX classes, or adding AUX classes which have MUSTs that are already in the NOT list.. Should there be language along the lines of "A DIT content rule SHALL NOT be added or updated in a way that would invalidate existing entries"?

Given that this means:
        Server implementations MUST ensure that schema is not
        updated in a manner which invalidates entries present
        in the DIT.

No, that should not be added as a) I don't believe implementations
do that and 2) it may be extremely difficult to implement.

If I recall correctly, there is language elsewhere in the TS which
suggests that existing entries may not conform to the subschema,
but if modified, must be made conformant by that modification.

That's the general rule I believe should be followed.

Kurt