Yes, you're right that the charset and natural language attributes
must appear first. That rule is near universal in modern protocols.
- Ira McDonald
High North Inc
Received: from zombi (zombi.eso.mc.xerox.com) by snorkel.eso.mc.xerox.com (4.1/XeroxClient-1.1)
id AA15628; Fri, 20 Mar 98 21:15:38 EST
Received: from alpha.xerox.com by zombi (4.1/SMI-4.1)
id AA10061; Fri, 20 Mar 98 21:09:42 EST
Received: from lists.underscore.com ([22.214.171.124]) by alpha.xerox.com with SMTP id <52148(4)>; Fri, 20 Mar 1998 18:09:39 PST
Received: from localhost (daemon@localhost) by lists.underscore.com (8.7.5/8.7.3) with SMTP id VAA25039 for <email@example.com>; Fri, 20 Mar 1998 21:06:15 -0500 (EST)
Received: by pwg.org (bulk_mailer v1.5); Fri, 20 Mar 1998 21:00:53 -0500
Received: (from daemon@localhost) by lists.underscore.com (8.7.5/8.7.3) id VAA24661 for ipp-outgoing; Fri, 20 Mar 1998 21:00:42 -0500 (EST)
X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0
Date: Fri, 20 Mar 1998 18:02:24 PST
From: Robert Herriot <robert.herriot@Eng.Sun.COM>
Subject: IPP>MOD: Problem in IPP encoding with "attributes-charset"
Content-Type: text/plain; charset="us-ascii"
I have found a problem that relates to the "attributes-charset" and
They must be in the first group of attributes (the operation attributes),
but they need not be first in the operation attributes group because the IPP
model document states that attributes in groups such as "operation
attributes" are unordered. Except for this case, that is good. But the
charset and natural language are special because they determine how name and
text values of all attributes should be interpreted. Such a text or name
attribute may precede the occurrence of "attributes-charset" and
This makes an implementation more difficult because it cannot convert the
octet string of the protocol to an internal String until it has found the
"attributes-charset" value which could come many attributes later.
We need to change the documents to state that two attributes need to be
first in the operations group, even though attributes in groups are
How are you other implementors coping with this?
By the way, XML doesn't have this problem because the charset comes at a
specific place at the very beginning and the language specification always
precedes the text which it applies to.