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

Re: Another Persistent Search question



Gordon Good wrote:
> 
> In the Netscape implementation, I'm fairly sure that an entry is only returned
> to the client if it matched the query after mods were applied.
> 
> So, for example, if the psearch used a filter of "foo=bar" and the modification
> changes the "foo" attribute from "bar" to "baz" then no entry would be returned.
> 
> However, for the delete operation, the entry that was deleted is returned.
> 
> To summarize:
> If optype is ADD, a result is returned to the client iff the new entry matches the psearch criteria
> If optype is MODIFY, a result is returned to the client iff the entry matches the psearch criteria after modifications have been applied
> If optype is DELETE, a result is returned to the client iff the entry matches the psearch criteria prior to removal
> If optype is MODRDN, a result is returned to the client iff the entry matches the psearch criteria after being renamed
> 
> That's from memory, but I'm pretty certain it reflects reality.

Yes, I agree.  This should be documented in the psearch document so
people who choose to use it or implement it do so the same way.  I will
add this to the psearch draft soon.


> 
> There's an inconsistency there: you can detect when an entry leaves the result
> set for delete operations, but not for modify or modrdn operations. I'm not sure
> it's worth fixing that in the psearch document, but it should definitely be
> addressed by the LCUP work (and I'll bet LCUP addresses this already).

It is being looked at.  See section 5.1 of
http://www.ietf.org/internet-drafts/draft-natkovich-ldap-lcup-02.txt

-- 
Mark Smith
Directory Product Development / Netscape