[odf-discuss] Response to Andy Updegrove's article about the Foundation

[robert_weir at us.ibm.com]

And we need to keep in mind that interoperability means different things 
to different people.

For documents in particular, if you define it to mean "looks exactly the 
same everywhere" then only PDF will be acceptable.  On the other hand, the 
strict form of a PDF makes it a poor choice for a format that you want to 
view on a cell phone as well as a desktop, and to use for further 
processing. 

An intriguing look at anti-complexity is the document format used by the 
Amazon Kindle e-book reader:  
http://forums.digitaltextplatform.com/dtpforums/entry.jspa?externalID=30&categoryID=11

It is a subset of HTML, a rather spartan subset.  But it allows the user 
to change font sizes as needed on this fixed display device, and the text 
simply reflows as needed.  No fixes page numbers,  no fixed layout, just 
the basics.  But it works.  We could probably do a very simple XSLT from 
ODF to this markup. 

HTML is someplace in the middle.  The fact that web designers need to 
carefully consider how their pages will look on different screen 
resolutions shows the additional complexity that the additional 
expressivity of HTML brings.

I've found that the most interoperable formats are those that specify 
everything related to appearance (like a PNG or similar bitmap) or those 
that specify nothing about appearance (like DocBook).  Those that live in 
the middle, like ODF in OOXML, have no end of troubles.  There is a reason 
why no one has created an office format with perfect interoperability yet. 
 This is an extremely hard technical task.  And quoting WTO regulations 
and anti-corporate rhetoric will not make that technical task any less 
challenging.

In any case, when looking at interop, the main categories of possible 
problems are:

1) Applications that don't support all of the standard (functional 
subsets)
2) Applications which implement features of the standard incorrectly
3) Applications which extend the standard in proprietary ways
4) Places where the standard is ambiguous
5) Places where the standard is silent
6) Places where the standard is incorrect
7) User errors

>From what I've seen by looking at ODF interop, the problems in practice 
are #1, #2 and #6.  I have not seen a lot of issues where interop would 
improve by changing the text of the ODF standard.  That isn't to say that 
there are not problems in the spec.  But in practice, those are not the 
main impediments to interop today.

Now you may wonder what #6 is all about.  What is a user error in this 
context.  Let me give you an example.  At our interop camp in Barcelona,we 
had different vendors create an ODF document in their editor to match a 
given PDF file.  Try to make it look the same.  One participant (an IBM 
executive whom I will not name) was good enough to join us, and used Lotus 
Symphony to create the document.  This document, when viewed by other 
editors was all messed up in the layout.  Was the problem with Symphony? 
ODF?  No.  The problem was this particular user did not know the features 
of the product well.  So, for example, when he needed to have  page header 
with centered and left aligned portions, he did this by inserting spaces 
until it looked good.  He didn't specify alignment.  He simulated 
alignment with spaces.  This may work fine when the document is loaded on 
this same machine, with same screen resolution, with this same editor. But 
it breaks when you look at the same document on another machine.  It is a 
very fragile technique.   I don't think we want ODF specified to the point 
where something like this is guaranteed to look the same everywhere.

So, I'm hoping this new Universal Interop Council takes a broad look at 
the problem in practice.   Certainly a careful look at the standards is in 
order.  But we also need a function like the Web Standards Group, to 
ensure that pressure is kept on implementations and vendors to implement 
the standard fully.  (How about an AcidTest for ODF?)  And there also 
needs to be some user education.  If you want to author a 
portable/interoperable document, then what should you do and not do?

Regards,


-Rob

odf-discuss-bounces at opendocumentfellowship.com wrote on 02/07/2008 
10:52:50 AM:

> Rob Weir:
> >> What would be interesting is an examination of standards or
> >> technologies that you believe have provided "universal
> >> interoperability".
> 
> Ian Lynch:
> > I doubt there is any absolutely universal interoperability standard 
but
> > ASCII for text characters
> 
> And then, it's only good for English and Latin :-)
> Try writing Fu?, ?brigens, sch?n or $BF|K\8l(B using ASCII.
> 
> Daniel.
> _______________________________________________
> odf-discuss mailing list
> odf-discuss at opendocumentfellowship.com
> http://lists.opendocumentfellowship.com/mailman/listinfo/odf-discuss
> 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.opendocumentfellowship.com/pipermail/odf-discuss/attachments/20080207/b0543afa/attachment-0001.htm