Re: funny article

<eric.kaun_at_gmail.com> wrote in message news:1102527502.957289.49990_at_c13g2000cwb.googlegroups.com...
> Jan Hidders wrote:
>> Costin Cozianu wrote:
>> >
>> > Do you mind if I ask you if this church of XML is abotu a data
> model or
>> > about a document model ?
>>
>> Hmm, I would say it's somewhere in between. The term "data exchange
>> model" describes it well, I think. It's in any case certainly not
>> something you should use for defining general conceptual models,
> (neith
>> is the relational model, but I digress) but in the near future it
> will
>> have all the necessary features of a data model.
>
> I'm still unclear about how a "data exchange model" differs from a data
> model

'cause it doesn't ;-)

> and while I have a general conceptual notion of a "general
> conceptual model," how does that relate to the aforementioned models?
>
> Maybe I missed this on an earlier thread. But in any event, although
> models can be re-commissioned for uses other than their original ones,
> I would think it significantly more difficult and risky than re-using a
> product.
>
>> But seriously, this is largely a matter of perspective. Yes, there
> are
>> grants and conferences are indeed sponsored. But to a very large
> extent
>> this has always been true in database research, and in fact most of
> that
>> money is coming from the same people as before, viz., the RDBMS
>> builders. This has never lead to any big problems before, and I don't
>
>> see why it should now.
>
> That's also largely a matter of perspective; SQL is a good
> counter-example, although it was led astray probably as much by bad
> internal decisions than by the addicts demanding their drug of choice.
> I'd say the vendors are trying to increase market share by jumping on
> XML after it acquired significant "mindshare" by virtue of its
> "coolness," not on any other basis.
>
>> Moreover, since when is it bad that scientists
>> are focusing on the real-world problems that industry says they are
>> having? I would consider that a feature, not a bug.
>
> Basic research is being neglected in all disciplines, though in
> programming the relatively small leap from incoherent concept to
> defective product makes this far less disciplined. I'd say scientists
> should focus on basic research and understanding, and that engineers
> should tackle real-life problems. This preference is based on results,
> not ivory-tower syndrome. And in any event, research on XML isn't based
> on problems industry was having - it's based on industry wanting to use
> XML.
I'm sure that is true of some research, but the origin of XML is very similar to the origins of pre-relational data models -- there was a problem to solve related to language-based data (compared to numbers, for example), so a variation on diagramming sentences was used to model propositions. So, unlike relational data models, XML, PICK, and other graph structures for propositions were pragmatic. They were never a solution looking for a problem. My research into PICK is pragmatic -- from the perspective of the data model (not from all perspectives) it seemed to work better than relational structures -- why? I suspect that some XML research is similarly motivated.

> So it's a focus on leveraging a solution, not solving a problem
> (except at a very low level); and I'd say that's the domain of
> engineering and "management," not science.
>
>> Finally, I can
>> assure you from personal experience that there is a very healthy
> debate
>> in the database research community about how to keep our academic
>> integrity. There are lots of researchers that will give you a really
>> funny look if you say that are going to work for Microsoft. :-)
>
> And yet MS continues to hire massive numbers of smart researchers. That
> scares me more than any other trend in the "software industry."
>
>> That there would be very little in terms of results is simply
> complete
>> nonsense. There have already been very interesting results in both
>> theoretical and practical areas, and at the moment this only seems to
> be
>> getting better.
>
> I'll take your word for it, but consider that getting a turd to
> levitate poses a number of very interesting theoretical and practical
> challenges to science, and yet at the end of the day you have floating
> feces.

I can see you have an open mind about it at least ;-)

> I actually don't think XML is as bad as a turd, but it sure ain't good.
> The processes that elevated it to some mutant variety of a "metalingua
> franca," though, is disturbing in the extreme.

Reviewing the politics and marketing of how change happens in any field shows up better stories than any soap opera. I've been trying to figure out how "relational theory" got canonized even though it set some aspects of computing back years in my opinion. I suspect that the elevation of XML to an unearned status is not unlike a similar elevation of relational theory a couple of decades ago.

>
>> Especially, there is some very interesting interaction
>> going in between the programming language community and the database
>> community which is leading to very new solutions and addressing
> problems
>> that weren't getting much attention in the past.
>
> That's a good thing, to be sure.

Yes. The question about full computer languages and "spec'd" information, whether specified in XML or SQL, for example, is really interesting to me. Check out http://www.jini.org/webinar/ for an intersting take on SOA (I'm a Jini fan in spite of its lack of XML-like buzz about it).

>> I've been doing some
>> research myself on query optimization and it is very interesting to
> see
>> how differently the two communities tend to approach the same
> problem.
>
>
>
>> > I mean a modest software engineer like me, even if I was
> brainwashed by
>> > XML propaganda, I wouldn't be able to use much in the way of
> building
>> > systems. That is after so many years of hype. That is contrast
> with,
>> > say, SQL, which inspite of the theoretical problems, still makes
> the
>> > world go round in many places.
>>
>> In terms of publishing formats, data exchange and file formats XML
> has
>> already made a lot of things much easier than they were and is in
> that
>> sense already making parts of the world go round right now.
>
> I don't think so - I think XML just happened to be a solution that was
> being pushed at the time that other technologies and various industries
> were pushing standardized integration, plus it had enough pedigree
> (*ML) to fend off the curmudgeons. It slipped in the back door of the
> club just as the doorman turned to take a leak in the alley, and before
> the heavy crowds hit.
>
>> But if the systems you build don't involve that,
>
> What, exactly, do you mean by "that"? "That" is the critical question.
>
>> then nobody in my church is
>> claiming that you should use XML anyway.
>
> Maybe not the church elders, but mobs of fecund, undereducated
> red-state reconstructionists (unrestrained by book-learnin') are.

>> From the database perspective one cannot really make a judgement
> until
>> XML as a data model is truly and well there, and things have only
> just
>> begun to take shape.
>
> Pardon me if I'll continue to be skeptical of a sorta-model that arose
> from document markup syntax and is trying to graft on structure,
> semantics, and queries after the fact. That strikes me as the
> antithesis of model.

XML, PICK, and other graph implementations might have started with pragmatics -- something that works -- so that the "theory" isn't completed at this point, but I might (someday) be bold enough to claim that the other way around -- starting with a theory and then trying to make it practical -- yields less favorable results. I'd rather start with something that works and have people in the back room trying to match a good theory to it than start with a theory and have people in the front room struggle to make it useful.

>> The essential component, a query language, is not
>> even a finished standard yet and query optimization research and
> other
>> typical database subjects (concurrency, integrity, et cetera) are
> still
>> very much in their infancy.
>
> What difference does query optimization really make to the model? Yes,
> I know there's a huge practical implication there, but a query language
> and query optimization do not a model make.

Agreed on the optimization side, but the model itself really has quite a bit to do with a query language, right?

> The apparent conceptual
> underpinnings of XML, with references to elements and nodes and paths
> and sections and such, are byzantine.

Agreed. This is where starting with something practical and then evolving it in whichever way is also not a good plan. If we had started with practical solutions, discovered a good theory (mathematical metaphor) that aligned with the solution, then start fresh and make a good practical solution with sound theory, that would have been better. The distinction between attributes and elements in XML is worse than the distinction between relations and scalar values.

> Unless I just haven't read the
> right intro to the topic... even Wadler's, while head and shoulders
> above many of the others, is horribly... ungrounded? Unjustified? It's
> much like reading an article of faith, at least to me.

Yup -- I wonder where they learned that "data model as religion" thing?

> - erk

Always enjoy reading you, erk. --dawn Received on Thu Dec 09 2004 - 03:01:28 CET