3 ms·
> A transpiling approach makes more sense to me. That's the solution we've been using, oft referred to as ORM, to work around the lack of this syntax. But then
by 9rx 2y ago
> A transpiling approach makes more sense to me.
That's the solution we've been using, oft referred to as ORM, to work around the lack of this syntax. But then you had to live in a world where you had an almost SQL, but not really, language that doesn't always quite map perfectly to SQL and all the pain to go along with that.
Now instead of that song and dance you can compose queries with plain SQL and simple string concatenation. That is a huge win, even if troubling that it has taken so long (and still not widely supported).
- jmull 2y agoGenerally, ORM's try to present a high-level abstraction of your data and data model than the engine does natively. That makes it the ORM's responsibility to implement the abstraction in a good way. Problems come in when the "good" way depends on factors the ORM doesn't consider or expose well. I actually mean transpiler in a mainly syntactic sense, and would want it to avoid higher-level abstractions (after all, as you point out, ORMs have that ground covered).
- 9rx 2y agoWhile that is closer to ORM in the classical sense, the modern use – and why I used 'oft referred to' – is more accurately a query builder, but get called ORM regardless. The reason people reach for this type of tool is because there hasn't been a good solution to this with SQL proper, necessitating another language. Until now.