Re: [opendocument-users] New question (2): Reference implementation?

[robert_weir@us.ibm.com]

Ian <ian.lynch@theingots.org> wrote on 05/17/2009 05:48:25 AM:

> 
> Is OpenOffice.org not a reference implementation? After all, odf came
> out of that project.
> 

It comes down to definitions.  Some people say a reference implementation 
is one that is co-developed with the standard and is a "proof of 
implementability" of the feature of the standard.  Others say that a 
reference implementation's code must be freely available.  Others say that 
it must pass a conformance test.  Others require that the reference 
implementation implements all of the features of the standard, including 
all of the optional features.

So, by some definitions, OpenOffice.org could be called a reference 
implementation.  My point was that neither OASIS nor the ODF TC has 
adopted a definition or designated a reference implementation, nor does 
our process require this.  What we do require is that a specification have 
three successful implementations which conform to the standard before we 
can vote on it as an OASIS Standard.  A single reference implementation is 
not part of our formal process.

That's my answer wearing my TC Chair hat. 

Wearing another hat, as an implementor of ODF, I'd say, yes, of course we 
look at OpenOffice and treat it as a reference implementation.  I assume 
all (or most) vendors recognize that it is the most-established and 
fullest implementation of the standard, that most ODF documents in 
existence were created in OpenOffice, and that if you want to have 
reasonable interoperability you must test your ODF output with OpenOffice.


-Rob