PC Week on "Why XML is Failing" - Comments from Megginson and DeRose
Date: 26 Apr 2000 14:56:19 -0400 From: David Megginson <firstname.lastname@example.org> To: email@example.com Subject: Re: PC Week on "Why XML is Failing" ?????
"Michael Champion" <Mike.Champion@softwareag-usa.com> writes:
> Take your favorite antacid and check out > http://www.zdnet.com/pcweek/stories/news/0,4153,2551691,00.html > > The essence of the article is that vertical industry B2B efforts are > flourishing but the "pure XML" efforts such as OASIS are not: > > "The losers are pie-in-the-sky XML-indoctrinated organizations, such > as the Microsoft-spearheaded BizTalk.org and OASIS. The automakers > chose to bypass BizTalk and OASIS to develop proprietary (but > published) schemata. Every big organization will do the same, and > though some of their XML work may feed back up to OASIS or BizTalk, > every one wants standards for their cliques only."
[Megginson:] I looked at the article, and it sounds about right (except for the headline and the conclusion) -- I never bet much on the success of schema repositories anyway.
In a sense, this is our own marketing hype coming back to haunt us. Two years ago, everyone was pushing XML like a tower of Babel that magically make all data formats compatible. That sounded a lot better than the truth -- that XML is a simple, common layer for representing tree structures in a character stream -- but we certainly deserve our share of headlines like this now.
Conclusion: XML is succeeding, but not by the criteria that many in the XML community originally set.
All the best,
David Megginson firstname.lastname@example.org http://www.megginson.com/
Date: Wed, 26 Apr 2000 15:29:40 -0400 From: Steve DeRose <Steven_DeRose@brown.edu> To: email@example.com Subject: Re: PC Week on "Why XML is Failing" ?????
At 1:17 PM -0400 4/26/00, Michael Champion wrote:
>Take your favorite antacid and check out >http://www.zdnet.com/pcweek/stories/news/0,4153,2551691,00.html > >The essence of the article is that vertical industry B2B efforts are >flourishing but the "pure XML" efforts such as OASIS are not: > >"The losers are pie-in-the-sky XML-indoctrinated organizations, such as the >Microsoft-spearheaded BizTalk.org and OASIS. The automakers chose to bypass >BizTalk and OASIS to develop proprietary (but published) schemata. Every big >organization will do the same, and though some of their XML work may feed >back up to OASIS or BizTalk, every one wants standards for their cliques >only." > [...]
[DeRose:] I agree they have some naive ideas in the article; but overall, it looks like old news to me. I'm glad the press has now figured out what many XML people have been saying for a long time: XML gets rid of totally pointless incompatibility over how to punctuate your data structures when you pass them around, but it doesn't solve the problem of disagreeing over the real data structurest that ought to be passed. That's just what was intended.
Expecting XML to solve that would be like expecting linguistic grammar to solve political conflicts. Without a basic (if perhaps imperfect) agreement on grammar and vocabulary, all bets are off. But once you can actually talk to each other, you can *start* actually working on whatever the real differences are. Common language is necessary, but not sufficient.
Now in natural language it's best if negotiators speak some common language, but we have workarounds: highly skilled translators can substitute; and no culture makes their language intentionally opaque, or holds it under non-disclosure (well, maybe the Auca). Even so, translators simply do not exist for many combinations of the world's 5000 or so languages (300 or so if you only count ones with armies). So, for example, the UN uses only a few "main" languages; even a mere 300 languages would lead to 900 potential translations -- and there are many more B2B, e-commerce, and other corporate content things than a mere 300 or 5000.
Such linguistic workarounds are infeasible in computing, so we need a common language even more: for which XML is working quite well. XML saves writing a parser for every app; it saves worrying about what to do with CR/LF, or what the 3rd bit over in the 'charformat' mask byte is; it saves worrying whether *this* RDB's comma-delimited export form escapes "," with backslash or by quoting -- all that mindless, useless junk that has nothing to do with your information, but causes a continuous drip of wasted programmer-years.
But XML does *not* solve the problem of semantic heterogeneity (a topic discussed in reams of database articles). XML never meant to; nor does it seem a feasible task. From my perspective, one of the best things I expected from XML was simply that it would make the *real* semantic problems visible: it is no longer so easy to miss the substantive differences behind a veil of syntactic obfuscation. By making the problems visible, at least we can deal with them consciously and directly.
Well, now we're there. XML lets us all see, in plain text and schemas, just what the differences between two competing systems were to begin with; and companies that want to interchange, convert, or compromise can do so explicitly and clearly. The process also helps directly: as companies and whole industries move to XML, they get a perfect chance to review and perhaps fix schema problems that have been there a long time.
Companies that are arch-rivals may avoid compromise, or even introduce different schemas just to create incompatibility; I don't think there's anything we can do about that. But there's a Darwinian process going on here, and being gratuitously different can lead to extinction. Working together is more efficient than working alone, and economies of scale favor standards in the long run.
At the same time, perhaps we will find that some of the mavericks have seen something everybody else missed, and we'll all be tweaking our schemas to add it.
Let the weeds grow along with the wheat; if we try to root them out now, we'll doubtless kill some of the good stuff too. They'll be weeded out in time.
Just saw Tim's posting while writing this one. I second his points too.
Chief Scientist, Scholarly Technology Group, and
Adjunct Associate Professor, Brown University
North American Editor, the Text Encoding Initiative
This is xml-dev, the mailing list for XML developers. List archives are available at http://xml.org/archives/xml-dev/
Prepared by Robin Cover for The XML Cover Pages archive.