Re: XML storing and management

On 27 sep, 22:52, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
> Jan Hidders wrote:
> > On 27 sep, 19:07, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
>
> >>Jan Hidders wrote:
>
> >>>On 27 sep, 16:27, Bob Badour <bbad..._at_pei.sympatico.ca> wrote:
>
> >>>>Jan Hidders wrote:
>
> >>>>>On 27 sep, 02:19, JOG <j..._at_cs.nott.ac.uk> wrote:
>
> >>>>>>Ok, so why is it exactly cdt, despite the inherent flaws of a
> >>>>>>hierarchical model such as XML, it has seen such widespread uptake?
>
> >>>>>It's all hype, of course.
>
> >>>>>Btw., what fundamental flaws?
>
> >>>>Well, let's see... How about we start with: "The inability to re-order
> >>>>the data without changing meaning and without destroying information." ?
>
> >>>"Hierarchical models such as XML" are not necessarily ordered-only
> >>>data models. In fact most proposals for semistructured data models
> >>>before XML weren't.
>
> >>>But even in XML this is not a big problem. Whether reordering destroys
> >>>information or not depends on your interpretation of the data. If you
> >>>send me an XML document and in addition tell me that certain parts
> >>>represent sets then I can reorder them without destroying any
> >>>information. The fact that XML is an ordered data model only implies
> >>>that it *might* destroy informaton, not that it *must*.
>
> >>That's a nit. If one cannot always safely reorder, then one cannot
> >>safely reorder.
>
> > I thought we were having a serious discussion, not playing trivial
> > word games. My mistake.
>
> I am having a serious discussion. I am not the one picking at nits.

I disagree. I think you are.

> What your position boils down to is: XML is needlessly complex.

Not really. What I said is that concerning the aspect we were discussing it is actually missing a construct. So my position is more accurately described as that it is "too simple", not "too complex".

> As a
> result of the needless complexity, one cannot re-order the data without
> changing meaning and without destroying information.

That is too imprecise to be correct. You can in some sense always reorder if you want to. What I said is that whether this loses information or not is a matter of interpretation. Note by the way that this is also true for the Relational Model: you cannot always arbitrarily permute the atomic values in a relation without risking changing its meaning. Also there it is a matter of interpretation whether this is actually a problem or not.

> BUT if we add even
> more complexity, we can sometimes re-order data. Sometimes.

Yes, when it is appropriate, which is not always.

• Jan Hidders