Maybe too simple

[htoug]

in reply to RFC: Simple DBI abstraction

I feel that it is just too simple.
It might work in an environment where you never have to do a join or in other ways really use the fact that you have a database system out there with a much greater potential.

But maybe I am predjudiced, I work in a place where our database (yes we only have one for all our data, although we do have copies for analysis, reporting etc, and a smaller version for development and testing) has over 1500 tables and is currently at ~50GB and growing.

I my experience, as soon as you grow beyond using the database as a way of just storing simple data, you will have joins in most of your queries, and so very little use for Class::DBI and related socalled simple tools.

The logical and the physical models are just too different! Your application needs data that is in a form that is suitable for its needs, the database needs the data to be normalised and easy to access for many different purposes.
Your 'return on investment' will be much greater then.

/me leaving soapbox and returning to lurkmode.

[Juerd]

I my experience, as soon as you grow beyond using the database as a way of just storing simple data, you will have joins in most of your queries, and so very little use for Class::DBI and related socalled simple tools.

Class::DBI is great if you see it as an SQL based object persistence framework. If you try to use it as a replacement for SQL itself, it suddenly becomes a whole lot less useful. It's not a querying language - in fact, that is what it tries very hard to not be :)

About so-called "simple" tools, I'm sure you will find that a simple interface can very well be used in complex designs, having looked at DBIx::Simple. It doesn't abstract, it simplifies. Please don't see one as the synonym of the other.