Re: [boost] Boost.SQL, SOCI and beyond

From: Nicola Musatti (Nicola.Musatti_at_[hidden])
Date: 2006-10-09 05:50:07

Steve Hutton <shutton <at> featurecomplete.com> writes:
[...]
> Well, back to Jeff's point about how it would be nice if the
> specification of the column names in the query was somehow validated
> against the the column names in the o/r mapping at compile time,
> to find any disconnects.
> 
> It's an interesting idea (might require full SQL parser to achieve?)

The way we do it is as follows: The representations of both classes and tables
are explicit in code, as is the description of how data members or accessors
are associated to fields. The code that performs queries is thus generic. We
separated the classes that are used for data exchange from the actual
application logic, so that the data exchange classes may be automatically
generated.

The tools we use for database design & maintenance can output the db structure
in SQL DDL, which we parse to generate the data exchange classes and those
part of the data exchange infrastructure that depend on the db structure. This
could obviously be done by connecting to the db and querying its structure,
but SQL source files are more convenient to store and handle with a version
control tool.

As our schema includes in excess of 250 tables this results in tens of
thousands of lines of code that are automatically maintained. This obviously
doesn't handle the case of the schema being altered in between builds, but as
we spell out in full the involved field names in all queries in general either
this isn't a problem or causes an error in the application.

Cheers,
Nicola Musatti