OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

office message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [office] what do mean by standard, and extensible?

First, "ODF" does not define what an open standard is.  A number of us, me included, regularly share ideas and thoughts and what this all means from a "Big Picture" standpoint, but these are just personal opinions and don't necessarily reflect the views of the ODF TC as a whole.  In the end ODF is exactly what is in the specification, no more, no less.

In any case, standardization and interoperability are two related but distinct questions.  It is possible to have a complete specification but still lack interoperable implementations.  And it is possible to have interoperable implementations even in the absence of a complete specification.    

Or look at it another way -- what causes interoperability problems?  It could be missing pieces in the specification.  But it could also be ambiguous specification, erroneous specification, implementation defects, partial implementations by design, etc.  When interoperability is achieved, it comes from cycles of compliance testing which lead to detecting and fixing defects in the specification, the implementations, as well as the test cases.

On the other hand, as we've seen with OpenOffice, sufficient effort on interoperability testing with MS Office, even in the absence of a public specification of Excel's formulas, has lead to a certain level of document interoperability.

Interoperability is a convergence of implementation behaviors that is fed from many information sources, including specifications, compliance testing, interoperability testing, etc.  The more sources of information, the better.  The ideal is an open standard, with a multi-vendor effort at compliance testing, which benefits the specification as well as the implementations.

As to extensibility, that is the X in XML.  No need to apologize for that.  Something to keep in mind is that  some statements in a markup specification act to constrain or limit what the application can express, while others statements act to coordinate how an application records what they did choose to express.  So, the HTML standard enumerating a fixed set of form input controls limits compliant implementations to those controls, while the same specification requiring  an RGB color model specifies the encoding, but doesn't limit which colors can be expressed (at least within the color gamut of CRT's).  

So there is some art to writing a specification as to what degree of constraint versus coordination one wishes to enforce.   Overconstrain and you end up with something inflexible and therefore less relevant.  I think the Word's "art borders" in OOXML can be held up as example of something which is overconstrained to the detriment of the specification  (http://www.robweir.com/blog/2006/07/game-of-zendo.html).  

Of course, having a specification which allows everything but mandates nothing won't work either.   The best we can do is aim for the sweet spot that does what most people want and then have extension points (coordinating statements) in the specification so that implementations do what they must.  



Rob Weir
Software Architect
Workplace, Portal and Collaboration Software
IBM Software Group

email: robert_weir@us.ibm.com
phone: 1-978-399-7122
blog: http://www.robweir.com/blog/

"Bruce D'Arcus" <bruce.darcus@OpenDocument.us> wrote on 09/03/2006 09:14:28 AM:

> An issue we've been dealing with in the metadata group, Rick Jelliffe  
> is dinging us a bit on some OpenFormula related wiki content:
> <http://www.oreillynet.com/xml/blog/2006/09/
> freaked_out_by_odfs_definition.html>
> For metadata, I actually thing Rick is a bit off-base that there need  
> be such a gulf between the two -- one can standardize the model without  
> standardizing the specific markup; standardize some of the markup/terms  
> without worrying about standardizing all of it -- but I think it's  
> worth clarifying the issue he raises as we move forward.
> When do we standardize something, and when do we leave a structure in  
> place that allows extension to happen without the involvement of the  
> TC?
> When, for example, should a formula extension be submitted for  
> inclusion in the spec, and when not? Is extension here just a mechanism  
> to fill the time gap between implementation and TC standardization?
> For metadata, I think the standards of inclusion would likely be  
> different, but we'd still want some statement on this.
> Bruce

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]