The Open XML standardization process is in its most interesting time at the moment: the various National Bodies are deliberating, the extreme anti-Open XML people have largely abandoned the sillier of their claims and tempered the rest, the moderate Open XML people are bringing up a lot of good issue (many of which seem reasonable and fixable to me), the Ecma side is having to seriously consider what kinds of changes they can live with, and national bodies have to look at whether particular issues identified are at the showstopper level of seriousness (not all flaws in a spec are showstoppers: some are better left to maintenance, of course. And few “showstoppers” are actually reasons to vote no in any case, in the ISO context.)

From my perspective, the lion’s share of problems will be solvable simply by appropriate clarification of the text (i.e. the normal ‘wordsmithing’ that goes on in a standard prior to its final acceptance. I presume that the intent of fast-tracking precludes changes that invalidate existing documents or require changes to semantics: what good would ISO PDF be, for example, if it did not reflect the pre-existing reality of PDF? But I do see scope for syntax additions, which MS would have to support as part of some service pack. It will be a test of their seriousness.)

I am due to speak to Bureau of Indian Standards today, largely on the same topics as in my recent blog on developing Principles for evaluating standards. (This was the same blog that various anti-Open XML ranters portrayed as being pretty far-out; in fact it is not even cutting edge…) Microsoft has flown me up here to talk to the local standards committee, on invitation from the Federation of Indian Chambers of Commerce; there is quite a feeling here that India does not need any artificial barriers to trade and that document standards help industry.

Talking to the standards committee was quite an honour and very interesting: many perceptive questions. A couple of times the questions morphed into a weird and confusing phase, where plain answers seemed to be ignored, nuanced answers derided, and previous statements re-cast with odd meaning shifts. The topic was on whether Open XML provides mappings to the old binary formats. Of course it doesn’t: it is an XML specification whose *goal* is to preserve all the *information* in previous formats, not a transformation specification. (Just like ODF does not specify a mapping to the Star Office binary formats, and neither should it.) I couldn’t understand why the question came up repeatedly, each time more confused, until I realized that the two people asking the questions came from experts from non-India companies with ODF-based rival products. Say no more…

The anti-Open XML campaign seems to be taking two tacks now. One is to say that whenever an issue that can be solved, for example editorially with clearer text or distinguishing statements, is found, rather than fixing it we should just abandon the whole thing, under some kind of iceburg principle. However, it is not reasonable to use mythical or imagined errors to stop a standard! Another tack is to say that ODF represents best practice, and therefore whenever Open XML does something different, it is deficient. Imaginatively, the two tacks even get combined, so that if there is anything that is not putative best practice, it is enough to stop the whole show.

Of course, I would imagine that National Bodies will merely look at how severe an issue is in its impact. An example is the date base formats for spreadsheets: the fact that formatting systems need to treat the first two months of 1900 as a special case (out by one) in one of the available numeric bases is indeed an oddity, but it is really an edge case of academic interest, not a showstopper. It is literally an edge case: dates only start in 1900 or 1904. (Which ODF allows too. If I recall correctly, SQL_DATE start from about 50 years earlier.) Now this could be solved simply by saying “Dates in January and February 1900 are possible but deprecated” or whatever. Instead, there is the throw out the baby with the bathwater gambit.

The trouble is that fast-tracked standards that standardize pre-existing technologies require a completely different set of ideas of “best practice” to blue-sky standards. For example, the standard for ISO PDF has to reflect the actuality of real-world PDF, and cannot go arbitrarily changing syntax or semantics: to do so would be to go against the point of a fast-tracked standard. The best-practice for a fast-tracked standard is that the spec clearly describes the technology under consideration warts and all, avoids unnecessary platform dependencies (which is not to say that it need not describe platform-specific features) and that it does so in a way that allows maintenance and subsequent improvements. The trouble with best-practice talk, is that what happens when a standard has 1,000 best practice items and 2 bad-practice items? Is that seriously enough for a “no” vote? This is why I am expect a lot of countries will not vote simple “yes” or “no” (which is what I presume Microsoft and IBM etc would prefer) but will vote “no with comments” (or “yes with comments”) in order to get improvemens. (This is problematic for countries who don’t want to send someone to a ballot resolution meeting of course.)