Re: [oic] profiles: ODF 1.2 or 1.1

[robert_weir@us.ibm.com]

"Hanssens Bart" <Bart.Hanssens@fedict.be> wrote on 02/18/2009 03:58:51 AM:

> 
> Now the first ODF 1.2 community draft became available, I'd like your
> opinion on the ODF version we should target when advising the ODF TC
> regarding profiles.
> 

I'm not sure I understand.  This TC has the ability, by our charter, to 
define and standardize profiles.  We're not limited to advising the ODF 
TC. 

When I drafted the charter for this TC, I had in mind profiles such as an 
ODF/A for archiving and a ODF/W profile for web-based editors, or 
ODF-Mobile for mobile devices.  Obviously there are other similar profiles 
that could be imagined.  In either case I think 90% of the effort would be 
the same for creating a profile based ODF 1.1 versus ODF 1.2.

But I think the key thing is to ask what vendor interest is there in 
implementing a profile at this point?  It probably is not worth creating a 
profile unless you have at least three parties who are willing to support 
it.

> I do believe we'll sometimes have to do some work at a low-level -
> sometimes even down to the element/attribute level (for example "Basic
> text processor profile may omit meta:document-statistics, but Full
> profile must support it"), instead of staying at the highest level. 
> 
> So, should we focus on creating proposals for ODF 1.2 profiles, and
> perhaps "back-port" it to 1.1 when resources are available ?
> 
> Or do it the other way around and start with ODF 1.1, given that it'll
> still take several months before 1.2 becomes an (OASIS) standard, and
> ODF 1.1 has fewer features / widely implemented ?
> 

My guess is you would want to base the profile on ODF 1.2, since that is 
what vendors will be moving on to over the next 12-months.  But this is 
hard to answer in the abstract.  I think we need to first ask "what 
profiles?".

Some profiles might be to increase ODF support within a particular 
platform, such as desktop editors versus web-based editors versus 
cell-phones.  Other profiles might be vertically-oriented, like a profile 
of ODF+XBRL or a profile for archiving.

From the interoperability standpoint, we have issues today even with 
heavy-weight desktop editors.  Profiles won't help us there.  So I think a 
healthy emphasis on the test case work will give us the best results. Once 
we've handled that domain, then profiles would allow us to enhance 
interoperability among other classes of implementations.  But we might not 
want to go there until we've made some deliverables on the core 
conformance and interoperability test suites.


-Rob