The correct solution is to abstract away the database from the developer. An ORM forces a square peg into a round hole.

using DAOs for database access does not preclude the usage of SQL abstraction layers and/or ORMs within that DAO.

Though taking the hardline "purity" approach of hand-coding all the marshaling of data to/from business objects into the DAO layer on a field-by-field basis, or writing an ad-hoc automation system for this task, basically reinvents what an ORM already does out of the box.

Ding ding ding. Just write functions that satisfy a need. Swap out from underneath as needed.

Surely there are much better ways to discuss the advantages of ORM over SQL than a bug report on github.

What are your recommendations?

The advantage is that you can easily ping other developers, comment, and reference other topics.

> The advantage is that you can easily ping other developers, comment, and reference other topics.

I didn't mean to insinuate that the discussion on github should not exist- it really should, and that's exactly what github is for. I'm just questioning the use of the page to trigger discussion on HN. Even the wikipedia page on ORM would have better content to consider.

EDIT: Unless gittip's potential use of an ORM is news-worthy, in which case I am very out of touch with hacker news. ORMs are as commonplace as databases now and have been for a while.