RE: [office] One strictly conforming document? - Raising the Floor, not the Ceiling?

["Dennis E. Hamilton" <dennis.hamilton@acm.org>]

Hi Bob,

This is probably a good place for me to make this comment more emphatically.

I find it distressing that we are distracted by a red herring considering
what someone might do with the allowance for foreign elements and attributes
and think by lowering the ceiling from where ODF 1.0/IS 26300/ODF 1.1 place
it is a blow for interoperability.

I say this because the *FLOOR* (the minimum level for conformance) of all
ODF specification is so low that there is no assurance, by specification, of
a minimum level of common conformance among implementations whatsoever.
(Although processors are obligated to accept schema-valid ODF documents,
they are not required to provide semantics for any of it and only chance and
good intentions have whatever semantics that are provided be more-or-less
compatible.)

What this says is that the only interoperability that is achieved at this
time is by (1) reliance on a common code base with OpenOffice.org and/or (2)
all implementers seeing what other implementations do (with some apparently
being more attentive to this than others).  So prudence and good intentions
by implementers is the only factor that gives rise to interoperable use of
ODF at this point in ODF history.  The specification and abstract concerns
about where the ceiling is has nothing to do with whatever interoperability
there is, and it seems likely to remain that way for the foreseeable future.

I think lowering the ceiling is going to accomplish zip, and I see no reason
to move it as a safeguard against a hazard that has not been experienced,
especially when it involves complex conformance statements that seem to
reduce interoperability even further by allowing more variations for which
there is no clarification in the specification (e.g., what is a text
document with no content.xml file and what momentous concern has us
recognize a single Math XML file as ODF?).

 - Dennis

PS: Producers that do not produce strictly conforming documents are
*already* defined to be conforming producers under a reasonably well-crafted
definition.  That part of the definition has been there since OASIS ODF 1.0
in 2005 and I must presume that the folks who put it that way though it
accomplished an useful purpose.    It might be valuable to say more about
what it means to have a (strictly) conforming document after any foreign
elements have been eliminated by the specified procedure, but the intension
of the statement seems pretty clear.  At the moment, there is no normative
strictly conformant document, and I also agree that there should be.

PPS: I think it is great for a procurement policy to require that a producer
make (or be configurable to make, or be defaulted to make)
strictly-conformant documents only.    That is, to be operated as a
strictly-conforming producer.  You can then work on what it takes for those
to actually provide usefully interoperable and interchangeable documents.
That is very valuable.  At the moment, there is no strictly-conforming
conformance in the ODF 1.0..1.1 pantheon, and it will be good to bake that
case into the specification.  But I see no reason to lower the ceiling that
is already there.  I do think it is of great value to provide another
ceiling that you feel more comfortable with for procurement purposes,
although I wonder how you qualify products with respect to the floor.  I
have my installation of OpenOffice.org 3.0 set to produce only ODF 1.1
documents for that reason as well, although there is no
strictly-conforming-only option at the moment.  It seems to me that far more
will be accomplished for interoperability by procurement requirements with
teeth in them than anything the ODF TC will provide in the short term, with
or without a ceiling that allows for foreign elements and attributes in a
particular way.

-----Original Message-----
From: Bob Jolliffe [mailto:bobjolliffe@gmail.com] 
Sent: Tuesday, February 03, 2009 05:03
To: ODF office
Subject: Re: [office] One strictly conforming document?

[ ... ]

The Department of Science and Technology, and the South African public
service as a whole, is a user of ODF 1.1 and we hope soon to become a
user of ODF 1.2.  For us the benefit of mandating compliance to a
standard is to clarify procurement processes and to open the field to
competing vendors around an agreed upon standard.

Anticipating (and encouraging) a heterogeneous mix of application
software, we would strongly support any moves in the direction of
stricter conformance requirements and equally discourage any moves
towards sanctioning loose conformance categories - either for
conforming documents or for conforming applications.  Put simply, if
we have deployed three different word-processing applications we
require that the output of those applications are in strict agreement
with one another.  I would not agree to dropping the requirement that
producers are available to produce strictly conforming documents.  If
a producer is not capable of producing strictly conforming documents
then it should not be called a conforming producer.  A nearly
conforming one perhaps.  The option of creating additional conformance
classes for producers serves no useful purpose for us.  In fact it can
only make our procurement processes more convoluted and increase the
risk of procuring systems which are less interoperable than they might
otherwise be.

[ ... ]