Re: LDAP Extension Style Guide, re interaction between controls

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 03:08 PM 8/16/00 -0700, Bruce Greenblatt wrote:
>Kurt,
>
>A few days ago, you wrote:
>
>>Just one comment on this comment... "when extending an operation though
>>definition of one or more controls, the specification SHOULD clearly
>>specify interactions with other, previously-defined, extensions
>>(using other controls).  The specification MAY NOT restrict further
>>extension of the operation by placement of additional, yet to
>>be defined, controls."
>
>I suddenly realize that I don't understand your last sentence.  Can you please clarify with an example?  Thanks,


First, let me state that I believe 2251 needs a clarification like
(needs much work, will note this for later LDAPbis discussions):
  "Controls SHOULD NOT be combined unless the semantics of the
  combination has been defined.  A server MAY ignore non-critical
  controls (even those it recognizes) to establish semantics
  of the operation (e.g. if two non-critical controls are combined,
  one or both may be ignored).  A server MUST NOT ignore any
  critical control (e.g. if one or more critical controls are
  combined with zero or more non-critical controls, the server MAY
  ignore any of the non-critical controls, but must perform the
  operation as defined by remaining controls.  If no semantics
  are defined for this remaining combination of controls, the
  server MUST return unavailableCriticalExtension"

Such a clarification would only require extensions to define
control combinations which make sense.  This would be consistent
with common practices.

Anyways, the last sentence was meant to indicate that any extension
may be further extended.  In particular, the specification should not
detail or restrict interactions with yet to defined controls.  If
such interactions are appropriate, the later defined control should
detail the semantics of the interaction.

It should be also noted that any attempt to say "This specification
SHALL NOT be updated by future specifications" is pointless
as the future specification can update that restriction away.

>>Just one comment on this comment... "when extending an operation though
>>definition of one or more controls, the specification SHOULD clearly
>>specify interactions with other, previously-defined, extensions
>>(using other controls).  The specification MAY NOT restrict further
>>extension of the operation by placement of additional, yet to
>>be defined, controls."