All,
I sent this email out earlier this month. I received no comments. I
propose that we adopt the second interpretation of the length restriction
issue and update the IIG accordingly. Does this need to be discussed at the
next PWG meeting or teleconference? There were issues raised by some people
from India that might also be discussed.
Pete
Peter Zehler
XEROX
Xerox Architecture Center
Email: Peter.Zehler at usa.xerox.com
Voice: (716) 265-8755
FAX: (716) 265-8792
US Mail: Peter Zehler
Xerox Corp.
800 Phillips Rd.
M/S 139-05A
Webster NY, 14580-9701
-----Original Message-----
From: Zehler, Peter
Sent: Thursday, January 06, 2000 12:54 PM
To: IPP Discussion List (E-mail)
Subject: MOD - Loose length restrictions definitions for
'xxxWithLanguage'
All,
I have had a few conversations in this area and can see that
the length specification for textWithLanguage and nameWithLanguage is a
little loose. The description below applies to name as well as text.
One interpretation comes from first reading the definition
for 'text'.
"A text attribute is an attribute whose value is a sequence
of zero or more characters encoded in a maximum of 1023 ('MAX') octets. "
One interpretation is that the "super type" 'text' is
limited to 1023 octets. Therefore, the total length of 'textWithLanguage'
is limited to 1023 octets. The 'language' portion of 'textWithLanguage' is
limited to 63 octets. The result is that the length of the 'language'
portion of 'textWithLanguage' "eats away" at the length of the 'text'
portion. The two error cases, with respect to length, are the length of
'language' exceeds 63 octets and the combined length of 'language' and
'text' exceeds 1023 octets.
Another interpretation is from the definition of
'textWithLanguage' itself. The 'textWithLanguage' definition contains:
"The 'textWithLanguage' attribute syntax is a compound
attribute syntax consisting of two parts: a 'textWithoutLanguage' part plus
an additional 'naturalLanguage' "
This verbiage together with the length definition of
'textWithoutLanguage of 1023 octets seems to imply a different conclusion.
The total length of an attribute value of type 'textWithLanguage' would be
1086. (This ignores the lengths types and attribute name portions of the
encoding) In this interpretation the 'language' portion does not "eat away"
at the length of the 'text' portion. The resulting two error cases, with
respect to length, are the length of 'language' exceeds 63 octets and the
length of 'text' exceeds 1023 octets
Which interpretation is correct?
I favor the second interpretation since it simplifies the
transformation between the two forms of 'text'.
The resolution should be clarified in the IIG and moved to
the MOD on the next iteration.
Pete
Peter Zehler
XEROX
Xerox Architecture Center
Email: Peter.Zehler at usa.xerox.com
Voice: (716) 265-8755
FAX: (716) 265-8792
US Mail: Peter Zehler
Xerox Corp.
800 Phillips Rd.
M/S 139-05A
Webster NY, 14580-9701