Sorry Shish, you're wrong about ODF 1.2
Try ODF 1.5 or ODF 2.0, maybe.
The metadata requirements for ODF 1.2 actually did include two way lossless translation capability. Unfortunately these features did not survive the final cut, and were not included in the April 2007 submission.
You might also want to check the February 23, 2007 metadata proposal from Florian Reuter. That also would have delivered the goods and perhaps put ODF that grand convergence category of usefulness across desktops, servers, devices and web systems currently the exclusive domains of MS-OOXML and CDF+. Florian had devised a means of using metadata to describe the presentation aspects of content and structural objects. Very revolutionary. And based on the simple notion that bold, font, margins etc. are simply metadata about content and style objects.
Where the train came off the track had to do with the concept of an XML ID means of linking metadata to content. Not that there was anything wrong with this mechanism. It's actually quite clever. What went wrong was that Sun insisted that only those elements approved and supported by OpenOffice would be allowed to make use of XML ID metadata. For independent developers, this is a serious constraint.
Because of this constraint, the metatdata sub committee started off with six elements supported by OOo that metadata could be appied to. IBM then came in and asked for eleven more elements having to do with charts and graphs. The OpenOffice crew decided they could support this, so in they went. Then an interesting question was posed, "How are independent developers supposed to submit elements for metadata consideration?"
One of the more important control points Sun insists on is that of commit rights and project managers. Only Sun employees have these rights and can hold these important positions..
The more important point is made by Marbux (below: ODF is an application specific format designed exactly for OpenOffice. While other applications might partially implement ODF, interoperability and successful ODF document exchange require the OOo code base!
From Marbux: This is the only article I've found to date where IBM (Heintzman) flat out says IBM wants changes in OOo licensing, more in line with the Eclipse and Apache licenses. See pg. 2. Significant because it feeds the meme that IBM's own ODF-based development goal is proprietary closed source built on the OOo code base, e.g., Symphony, et cet.
And that has huge signficance once you realize that ODF is not the real standard; the OOo code base is the real ODF standard. Look around the world and you see that ODF adoption decisions by governments are all in reality decisions to go with StarOffice, OOo, or OOo clones. I haven't, for example, seen a single instance where a government decided to ride with KOffice. Why would they, with the interop issues between KOffice and OOo? The fact that OOo's code base is the real ODF standard will figure strongly in the comments. Couple it with Sun's iron-fisted control of the OOo code base, and you have vendor lock-in with a Microsoft partner.
But with 70 developers committed in China, where developers salaries are inexpensive, IBM will soon be in a position to threaten to fork the OOo code base using proprietary extensions. Is that their real tactic to force changes in licensing and gov