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

RE: LBER (BER restrictions)




Jim,

I would recommend at this point ot leave the text alone.  I would agree that specifications for controls MUST specify what is held in the value.  If the values are themselves BER-encoded, so be it.  This is the "common practice", but I don't think that it must be required.

Regards,
Tim Hahn

Internet: hahnt@us.ibm.com
Internal: Timothy Hahn/Endicott/IBM@IBMUS or IBMUSM00(HAHNT)
phone: 607.752.6388     tie-line: 8/852.6388
fax: 607.752.3681

Sent by:        owner-ietf-ldapbis@OpenLDAP.org

To:        <Ron.Ramsay@ca.com>, <mcs@netscape.com>
cc:        <ietf-ldapbis@OpenLDAP.org>
Subject:        RE: LBER (BER restrictions)




I wasn't thinking on my choice for AuthenticationChoice.simple  (should have been no). I Agree LDAPString is a no as well. I revised the table  below, and I'm back to the original question: Do the restrictions apply to  extension data (extended req/resp, and controls)?
 
If the text says that "ASN.1 values carried opaquely  as OCTET STRING values are not subject to the restriction", Does it  sufficiently describe what to send and expect from extension data? I suppose if  that data is described as an ASN.1 value in its specification, it  does.
 
Hmm, now I'm inclined to leave the text alone. Or maybe add  something saying that for controls and extended op's, those specifications  dictate what can/can't be in the value. Does anyone think that is  needed?
 
Jim

>>> "Ramsay, Ron"  <Ron.Ramsay@ca.com> 11/22/00 6:28:07 PM >>>
Whoa! The  exemption applies to ASN.1 values carried opaquely in the protocol
as OCTET  STRINGs. This eliminates LDAPString, LDAPOID,
AuthenticationChoice.simple.  Also, I don't think the format of the contents
of Control.controlValue,  SaslCredentials.credentials,
ExtendedResponse.response are specified - they  may be text.

The text should just say that ASN.1 values carried opaquely  as octet strings
are not subject to the  restriction.

Ron.

-----Original Message-----
From: Jim  Sermersheim [mailto:JIMSE@novell.com]
Sent: Thursday,  23 November 2000 9:57
To: mcs@netscape.com
Cc:  ietf-ldapbis@OpenLDAP.org
Subject: Re: LBER (BER  restrictions)


Looking at this again, I suggest we enumerate all  things that are OCTET
STRING values and explicitly state those that are  exceptions to the  BER
restrictions.

value                          exempt?

LDAPString                     no  
LDAPOID                        no
AttributeValue                 yes
AssertionValue                 yes
Control.controlValue           no?
AuthenticationChoice.simple    no
SaslCredentials.credentials    yes
BindResponse.serverSaslCred    yes
ExtendedRequest.requestValue   no?
ExtendedResponse.response     no?

I'm not sure  about those with ? marks.


>>> Mark Smith  <mcs@netscape.com> 11/22/00 1:44:34 PM >>>
Jim Sermersheim  wrote:
>
> In section 5.1 of RFC 2251, after enumerating the  restrictions on BER it
says:
>
> "These restrictions do not  apply to ASN.1 types encapsulated inside of
> OCTET STRING values, such as  attribute values, unless otherwise noted."
>
> I think this should  also apply control values and extended operation
values.
> If so, I'll  call those out as well here for clarification.

Are you suggesting that  Control/controlValue,
ExtendedRequest/requestValue, and  ExtendedResponse/response NOT be
subject to the BER restrictions outlined in  section 5.1?  I would prefer
that they continue to be subject to the  restrictions.  Why?  Because
controls and extensions can be  standardized and there are advantages in
having the restrictions in place  (both in terms of increased efficiency
and in terms of reduced  complexity).

--
Mark Smith
Directory Product Development /  Netscape