DBIx::Class and Class::DBI are the same kind of thing – ORMs that layer an OO abstraction over the database. They try to create a model for you with minimal effort on your side, but in so doing they keep very close to the structure of the database, and yet make it difficult to formulate non-trivial queries.

I prefer to write my model from scratch, talking SQL to the database directly. DBIx::Simple does not add any abstractions over using DBI directly – it just puts a cleaner interface on top of it, so you get to keep your sanity.

Makeshifts last the longest.

DBIx::Simple does not add any abstractions over using DBI directly

Actually, it does. Instead of working with statement handles, they have been abstracted away and wrapped in result set objects. There are some fundamental (subtle but very natural) differences: you wouldn't re-execute a result set, for example. It does stick to DBI's iterative approach to fetching, even though many would expect to be able to re-use a result set object and get the same data from it twice. This has not been abstracted, because it would harm performance immensely, especially with databases that don't support rewinding cursors natively.

Juerd # { site => 'juerd.nl', do_not_use => 'spamtrap', perl6_server => 'feather' }