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

ModifyRequest OPTIONAL PartialAttribute (Was: protocol-27 comments)



At 11:31 PM 10/28/2004, Jim Sermersheim wrote:
>>> 4.6. Modify Operation
>>
>>> ModifyRequest ::= [APPLICATION 6] SEQUENCE {
>>> object LDAPDN,
>>> changes SEQUENCE OF change SEQUENCE {
>>> operation ENUMERATED {
>>> add (0),
>>> delete (1),
>>> replace (2) },
>>> modification PartialAttribute } }
>>
>>I wonder if the PartialAttribute should be OPTIONAL in the grammar and
>>instead "REQUIRED unless otherwise specified" in the text, anticipating
>>new ModifyRequest operation types (per [LDAPIANA]) that need a different
>>data structure than PartialAttribute. Such operations can then add
>>trailing sequence elements (which servers are already required to
>>accept) at the end of the 'change' SEQUENCE.
>PartialAttribute allows zero elements (in the ASN.1 grammar), so an empty one may be specified. If I remember correctly, we decided this was good enough for purposes of future extension. 

While I couldn't find the discussion in the archives
(could have been off-list), I do recall consideration
of this issue.   IIRC, the suggestion was to replace
the components of a change with a CHOICE in a manner
which would have a BER-compatible encoding.  Anyways,
I think the conclusion of those discussions was that
Modify extensions were not overly constrained by the
current ASN.1 data type definition.

While I believe it unlikely that WG consensus would
support making the modification component OPTIONAL, I
will defer declaring consensus for a bit to allow
on list discussion.  Those who support making this
change should offer compelling reasons why it should
be made.

Kurt