Antw: Re: Q: accesslog, cn=subschema attributes

["Ulrich Windl" <Ulrich.Windl@rz.uni-regensburg.de>]

>>> Michael StrÃder<michael@stroeder.com> schrieb am 25.07.2014 um 18:37 in
Nachricht <53D287C4.9000505@stroeder.com>:
> Ulrich Windl wrote:
>> 1) Is it true that reqStart and ReqEnd use pseudotime for fractional
>> seconds instead of real time?
> 
> Yes.
> 
>> 2) When I query cn=subschema, the schema descriptions seem to be flagged
as
>> "operational attributes". Is this conforming to the standard?
> 
> Yes. (Why not?)
> 
>> If I query
>> operational attributes, I also see (besides structuralObjectClass,
>> createTimestamp, modifyTimestamp, etc.) objectClasses, matchingRules,
>> matchingRuleUse, attributeTypes, ldapSyntaxes, ...
> 
> Query what? Be more exact.

Talking about "* +"...

> 
>> A user querying cn=subschema usually wants to get actual schema
attributes.
> 
> What does "actual schema attributes" mean?

Well cn=subschema's "data" (normal attributes) are intended to describe (among
others) the classes and attributes. As you have to explicitly query
cn=subschema, nobody will see the "operational attributes" of cn=subschema by
mistake.  If the query specifies operational attributes, the user gets a mix of
true operational attributes and those that make up the schema descriptions. I
cannot see much sense in defining that.

> 
>> Having all as operational attributes seems strange at least...
> 
> I don't understand the issue.
> 
> From RFC 4512:
> 
>    A usage of directoryOperation, distributedOperation, or dSAOperation
>    indicates that attributes of this type represent operational and/or
>    administrative information.  That is, they are operational
>    attributes.

>From the theoretical aspect it makes sense as all schema definitions are used
to manage the DIT, but it's also data the user MUST know to change (or add) any
attributes. In my personal feeling I think that an entry describing operational
attributes should return the data (operational attributes) as normal attributes
while the operational entries all entries have should remain to be operational
attributes. Well you don't always understand the work of commitees...

Regards,
Ulrich