Re: [oiic-formation-discuss] Interoperability versus Conformity

[robert_weir@us.ibm.com]

"David Gerard" <dgerard@gmail.com>
wrote on 06/11/2008 09:53:33 AM:

> 2008/6/11  <robert_weir@us.ibm.com>:
> 
> > From the study I've done of the topic, the greatest source of
visual
> > interoperability problems today, with the ODF implementations
out there, is
> > not from any defect in the ODF standard.  It is caused by
incomplete/partial
> > implementations of the standard, where a particular feature is
implemented
> > partially, or not implemented at all.  With the way document
layout works, a
> > small failure in a single feature can have a global effect in
the document,
> > shifting lines, pages, figures, around.  Small failures
can make a large
> > difference.  This is a kind of problem that lends itself
well to testing, to
> > profiling and to working with vendors toward improvements.
> 
> 
> Do you or someone have a list of these? Sounds like an excellent start
> on an ODF Acid Test. And, of course, a good checklist of what features
> an ODF writer needs to implement in what order.
> 

We didn't have a formal report-out from that workshop,
but I can point you to the materials we used. 

Here's the link:  http://marketing.openoffice.org/ooocon2007/programme/odf_camp.html

We had participation from OpenOffice, Symphony, KOffice,
Sept Solutions (Symbian implementation of ODF viewer), AbiWord, Google
Docs, even MS Word, via the CleverAge plugin (with Novell's assistance).

As homework, each participant was given the same 4
PDF documents, each one illustrating a typical modern business document.
 Each of the documents had a particular point that made them interest,
such as use of form fields, or mathematical equations, or interactions
between numbered lists and tables.  But the examples were not especially
crafted to be cutting edge.  We weren't making an Acid test where
we expected everyone to fail.  We wanted to see some success as well
as room for improvement.

Each participant took the PDF file and then recreated
it in their ODF editor, from scratch.  At the workshop we then exchanged
all of the documents and took turns loading them in each word processor.
 So for example, Sun went up to the front of the room, showed us their
version of document-1 and then loaded the versions created by each of the
other word processors, document-1-ibm, document-1-koffice, document-1-google,
etc.  We tried all the interoperability paths we had time for.

The exact data collected then is now out of date (then
was OpenOffice 2.3 and KOffice 1.6 and Symphony beta 1.  Now we're
at OpenOffice 3.0 beta, KOffice 2.0 beta, and Symphony 1.0 has recently
been released), but I think the general observations are still reasonable:

1) Biggest cause of interoperability problems was
incomplete implementations of the ODF standard.

2) A large contributing factor was user behavior.
 This may sound odd, but indeed there are certain things a document
author can do that will make interoperability near impossible.  For
example one of the sample documents required a page header with a right
aligned page number.  One of the documents was prepared by someone
who, instead of using the alignment feature (he couldn't find it in the
menu, decided to mimic alignment by adding 50 or so spaces to put the page
number visually where you thought it look good.  But when loaded in
other editors, the page number was misplaced, sometimes wrapping over to
the next line. 

We've had some preliminary discussions on whether
we want to hold a similar event at OpenOffice.org Conference 2008, in Beijing,
and it seems likely.

Now there is a relationship between that kind of workshop
and our present efforts.  I'd much rather have the implementors of
ODF gather in Beijing and be focusing their effort on evaluating their
implementations against a draft version of the "OASIS Conformance
Assessment for ODF 1.1" than playing around with 4 documents with
no particular significance or traceability to the standard.

-Rob