Cart
[Home] [By Thread] [By Date] [Recent Entries]
This and the examples Uche presents are precisely why the ancients of SGML stayed out of programming languages. There was even an organizational separation at the ISO levels. Charles was marvelously scrupulous about that even when the youngsters around him (like me and a few others who now pass for old guys) were clamoring for it. Hungry for it. Ambitious for it. He quietly held us under the water until we relented or turned blue. When we were doing the MID design, I passed the first version (C++ in pointy brackets) to Steve deRose. He sent back an interesting critique that came down to "why code in pointy brackets?". Later, Charles and Yuri caught me outside the HyTime conference hall in Vancouver to ask esssentially the same question? It WAS doable; the MID designs worked. The question was, what did it bring to the standards table we didn't already have, and would it be anymore effective than what we had? When HTML began to replicate the features of the MID inside comments, I had to wonder if it wasn't all simply "this wins, that loses, who sez? who cares?" and at the application layer, that is essentially the case. Applications are products. But at the layer of the standard, it is wise not to add functionality for the advantage of one particular application that makes it hard or expensive for others to do their jobs. Eliminating options without regard to the commmon opportunity is usually a very bad practice. XML, like it's parent, is smart ASCII. Keep it that way. When building an XML application, all bets are off unless you claim to conform to another standard or specification, say DOM. Is XML post-OO? No. It is pre-LISP. Len http://www.mp3.com/LenBullard Ekam sat.h, Vipraah bahudhaa vadanti. Daamyata. Datta. Dayadhvam.h -----Original Message----- From: Al B. Snell [mailto:alaric@a...] Sent: Wednesday, May 23, 2001 10:45 AM To: Bullard, Claude L (Len) Cc: Eric Bohlman; Michael Brennan; xml-dev@l... Subject: RE: XML is _post_ OO On Wed, 23 May 2001, Bullard, Claude L (Len) wrote: > However, the interface might be (and very often is) > as simple as Import<->Export fields of hopefully > structured data. Can one code smarter interfaces? > Sure. Can one afford to do that often? No. So > very often it comes down to, show me your names, > descriptions, and field lengths, I'll show you > mine. Whatever we determine is common, we exchange. > Beyond that, the information loses or Mr Customer > pays one of us to make adjustments. Indeed; standards happen when commonly needed interfaces that really do the same thing compete and a winner emerges, often aided by pressure from a nice standards body or nasty vendor... ...but still standards happen. I am perplexed by people who say that agreed interfaces cannot happen on the Internet.... read the RFCs! People are not forced to conform, and many implementations add vendor extensions and all, but unless the standard is broken or vendors get nasty, there is usually interoperability as per the RFC - the vendor extensions don't prevent that... > Len ABS
|
