Lists Home |
Date Index |
On Tuesday, October 22, 2002, at 01:45 AM, Paul Prescod wrote:
> XML tools are typically designed with a "view" mindset. The XML
> document is the interchange format of some data.
Perhaps more accurately, *an* interchange format of some data.
> There is almost always logic, whether declarative or procedural,
> mapping to the persistence layer and the toolkits are designed to help
> with that. I wouldn't really no how to go about creating a
> SQL-queryable "view" in C as an abstraction over a variety of SQL, OO
> and legacy data sources. Is there (for example) an open source project
> or .NET component that would help me with that? Would I have to become
> a deep magic expert at Postgres or SQL server versus spending a half
> day with Expat?
No, you could write a check. Maybe read a book on n-tier architectures.
There are plenty more for all sorts of languages. Many of them come
with lovely little point and click modeling tools that let you design
the view and the mapping down to the physical database layout and do
nice things like make many to many mapping tables invisible to the
> My understanding is that Microsoft has tried various experiments about
> the "universal data view architecture" (ODBC, ADO, ec. etc.) and
> so-far, XML has "won". In my experience, it is quite uncommon to
> create SQL views outside of a SQL database, and very common (and easy)
> to create XML views of all sorts of information.
ODBC is one of the lower level layers in portable O/R mapping. All it
does is make different vendor api's look (mostly) the same. O/R
mapping libraries allow you to create views on top of this layer that
are expressed in objects. When fetching objects, you get both the data
and interesting behavior - rather than completely passive data like you
get with xml.
It makes factoring the business logic into reusable components very