> >Carl and others, including Keith Moore, have tried to express
> >that the NL/NLO
> >scheme is unduly complex and prone to error. Carl's proposal
> >represents a
> >simpler scheme where every text and name attribute would have
> >an explicit
> >natural language thereby simplifying the implementation with
> >fewer attribute
> >syntax's, and reducing the number of attributes which have
> >multiple syntax's -
> >all with NO LOSS of functionality.
>> Exactly. NLO 4 of 4 proposes exactly that: Drop textWithoutLanguage and
> and always use textWithLanguage and nameWIthLanguage in requests and
> responses. But 4 of 4 (unlike Carl's proposal) does not change the
> "attributes-natural-language" Operation attribute in requests and keeps it
> in responses as well for compatibility (though its not used).
>I disagree strongly with keeping the (unused) "attributes-natural-language" Operation attribute in responses. If it's not necessary let's apply Occam's Razor and get rid of it. Useless appendages just cause confusion for implementors. For example, what value should be sent? How does the client check the received value?
I'd go along with leaving it in as a MAY, purely for historical reasons, with a note to the effect that it's there only for compatibility and it is to be ignored.
See the original message at http://www.egroups.com/list/ipp/?start=4717
Free e-mail group hosting at http://www.eGroups.com/