Slashdot Mirror
Patent Threats In OOXML
An anonymous reader notes an initiative by the New Zealand Open Source Society to weigh in on the question of standardizing Microsoft's OOXML. The organization has authored a white paper (available in several formats, HTML here) laying out the ways in which the OOXML spec falls short of what a standard should be. From the article: "'If OOXML goes through as an ISO standard, the IT industry, government and business will [be] encumbered with a 6,000-page specification peppered with potential patent liabilities' said New Zealand OSS President Don Christie. 'Alarm bells are going off in many parts of the world over OOXML. Normally ISO draft standards would be drawn up by a number of stakeholder organizations, involving an often slow process of consensus building and knowledge sharing. Since many aspects of the office document format remain proprietary, OOXML has not taken this development track.'"
" It is already clear (from other /. stories) that the OOXML architecture seems rather shoddy and looks like something that was quickly put together. MSFT is trying to force it through iso rather thanb let OOXML succeed through its own merit... that alone draws suspicion to the quality of OOXML."
/. and Rob Weird, why don't you read the spec yourself, or just go to http://openxmldeveloper.org/ and read some sample code and articles, and then decide if it's a "shoddy" architecture. And it wasn't quickly put together, it was put together over a number of years, and the ECMA process itself took over a year.
Oh please.
Rather than relying on FUD spread by
-- "I never gave these stories much credence." - HAL 9000
OOXML could one day be called a standard, but it will never actually be one. Firstly, it's not a specification for a program to be written to, it's an after the fact description of MS hodge podge code.
At 6000+ pages, nobody will ever truly understand all of it. It's not humanly impossible, but who wants to make understanding a single 6000 page document their career? Consider, if you can read 100 pages with full comprehension per work day, it'll take over 3 months just to get through it.
With 6000+ pages of crap to implement, compliance testing will likely never reveal all of the subtle incompatabilities between different implementations even with an extensive bakeoff. If MS were forced to re-implement from scratch, I doubt even they could fully comply with their own spec. This is, of course, their true intent. They want an "open" specification so complex and screwy that there can be one and only one compliant implementation (theirs, of course).
Given the massive amount of up-front overhead to implement, there will likely be ONE attempt that will never be re-written no matter how much it cries out to be re-written.
Successful standards tend to be under 100 pages long and comprehensible in an afternoon or at least can be logically sub-divided into independantly useful subsets that meet those criteria. That's the whole reason TCP/IP has been a success. ODF pushes and probably exceeds the limit by a bit but at 6000+ pages OOXML is a non-starter.
Truly successful standards can be trivially shown not to be a patent minefield. Either by being simple enough to fully analyse and research or due to a simple and comprehensive blanket coverage. I hate to even imagine the legal bills that would result from handing a legal team a 6000 page document with instructions to identify all relevant patents within.
Any standards body that would ratify such a thing is fully incompetant at its core function and should be ignored until it goes away. The fact that OOXML is still alive and kicking raises big questions about ISO in my mind. Their final answer to OOXML will go a long way to answering those questions. Of course, the ISO2000 and 2001 boondoggles already started that questioning.