Lists Home |
Date Index |
>> You asked how XML is an improvement over SQL. I told you.
> I must have missed it. I don't see an improvement. Its a request
> written in a syntax that is applied to a store and gets you some passive
> data in a particular structure. Whether the thing is rows and columns
> or nodes and arcs doesn't seem too relevant. Some amount of the store's
> structure is revealed in the query and thus leakage occurs regardless.
No, the structure of the request and the response is generally
completely unrelated to the underlying structure of the data store. In
fact, the structure of the request and the response is ideally an
international standard like SVG, RSS, HRML, ebXML etc. This structure is
unrelated to the databases you use to structure it.
>> Now you bring in object-relational mapping as if I've never heard of
>> it. If you wanted to know how XML is different than an object view of
>> a relational database you should have asked that question, now how XML
>> is different than SQL.
> I know how its different. I don't get why its any better. Because
> whether you say XQL or SQL its still a QL on passive data.
I didn't mention XQL. Yes, I think that XQL has many of the same
limitations of SQL. I can see a class of applications for it, but it is
much smaller than the set of applications for XML in general.
> You'd never know it from the rhetoric. So why passive? What was wrong
> with intelligent rather than dumb information?
The basic idea of an object is to wrap up some data with some methods
(behaviour). The basic presumption is that the person providing the data
knows enough about what the person using it wants to do to provide the
correct and complete behaviours. But typically in networked applications
that is not the case. It is better to just separate the data from the
behaviour and let the client associate whatever behaviours it wants. A
simple example is different display semantics for different devices. If
you, as a producer, really have behaviour you think the client will want
then you can embed it in a <script> tag or point to it somehow. But the
data and behaviour should be still separate.
Plus there are many technical issues with passing around mobile objects:
denial of service, analysis is difficult because of Turing completeness
issues, "secure" virtual machines usually have bugs, cache semantics are
really tricky, etc. etc.
> ... Because I've been at
> this long enough to recognize a circle closing when I see it and I think
> things are moving in the wrong direction.
I don't see a circle closing. XML is, as far as I know, the first
universally used information interchange format. It certainly is not
analogous to either Smalltalk or SQL.