[pgt]

Most engineers who say they want an ORM, really want query composition.

[Aperocky]

Not really, they (I) wanted `insert(Item)` that just inserted the item into the table.

I don't really care about the query, just that it's inserted correctly, along with other operations that an ORM provide (get/delete/update,etc)

[icedchai]

I think the data mapping part is pretty important: a programmatic way to map from SQL result sets to objects. That stuff is incredibly tedious (and error prone) when you do it manually.

[pgt]

Decomposing the result set is much simpler if the caller composed the query. Harder if you have to parse SQL to know what to expect back.

[icedchai]

It depends. I've written a few half baked object mappers and never needed to parse SQL, since I've always used metadata from the result set (column names, data types, etc.)

[yunohn]

What’s a good solution for that without a classic ORM?

[nicoburns]

Conditional text interpolation with named parameter bindings works quite well. We’ve basically abandoned the ORM for select queries at work as we find this approach more readable.

[yunohn]

That sounds… flakey. Variable interpolated query strings are not even close to a substitute for an ORM?

[pgt]

Results are easily decomposed if you composed the query.