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

Re: I-D ACTION:draft-zeilenga-ldapbis-rfc2253-02.txt



At 08:04 AM 1/2/2003, Hallvard B Furuseth wrote:
>2.3 (Converting AttributeTypeAndValue) says:
>
>  If the AttributeType has a registered short name [REGISTRY], that
>  registered short name, a <descr>, is used.
>
>I think the short name "SHOULD" be used,

It is.  Section 2: "This section defines the RECOMMENDED algorithm..."

>and MUST be used only if it is in the table in 2.3 in RFC2253.

Section 3 mandates that implementations recognize names in
the table.  Is there a need to mandate their use?

Considering that some implementations took the table as being
"an example" and as DC is not in [4], there actually be implementations
generating "0.9.2342.19200300.100.1.25=foo,0.9.2342.19200300.100.1.25=com"
(or the equivalent using the #ber-hex notation).  Are these
implementations to be outlawed?

>That retains compatibiblity with
>implementations that followed RFC2253 and used the dotted-decimal
>form for other attribute types.

I think we need to continue to allow use of numericoid instead
of descr.

Kurt