I think that we are in agreement per my previous email
except for the error message.
An error message that says bad length is more useful that
'client-error-bad-request',
but if we do this for uri's, why don't we do it for other types,
especially text and name
types where the length can also be exceeded easily?
Bob Herriot
At 12:11 PM 5/11/98 , Carl Kugler wrote:
>P.S.=A0 My preference would be to always return
>'client-error-request-uri-too-long' for any 'uri' that fails the
length check.
>That way, the checking can be done in a lower layer that understands
attributes
>syntaxes but not specific=A0 attributes.
>
>-----------------------------------------------------------------------<=
br>
>
>If the case of an unknown or unsupported attribute with 'uri' syntax
that fails
>the length (<=3D 1023) check, does the Printer return
>'client-error-request-uri-too-long' or 'client-error-bad-request'?=A0
Reading
>section 15.3, I get the impression that the
'client-error-request-uri-too-long'
>only applies to the "document-uri" attribute.=A0 And in the
particular case of
>
>unknown or unsupported attribute
>IF the attribute syntax supplied by the client is supported but the
length is
>not legal for that attribute syntax, REJECT/RETURN
'client-error-bad-request'.
>
>=A0 -Carl
>