Fork me on GitHub
#sql
<
2018-12-07
>
souenzzo19:12:35

Talking about "jdbc-next" and ddl, you have already seen the #walkable ?

souenzzo19:12:24

I think that #walkable is a awesome idea and "connects" with a lot of ideas of "idiomatic clojure"

seancorfield19:12:47

Walkable is certainly interesting but I think it's kind of a compromise between Datalog and several SQL DSL libraries that already exist out there -- and it's kind of verbose for some things that are simple in SQL (esp. the whole floor plan thing that's basically a redundant schema definition -- redundant insofar as you're basically duplicating/adding to the information that your DB already has).

seancorfield19:12:31

And Walkable is built on top of clojure.java.jdbc already.

seancorfield19:12:10

I think it's great that folks are building all these things on top of java.jdbc. I don't like ORMs, I find Datalog-style queries hard to read (because I'm very used to SQL), I don't like my SQL externalized (e.g., HugSQL), but it's important that those things exist for folks who want to use them.

seancorfield19:12:50

clojure.java.jdbc "2.0" needs to retain the low-level foundational work -- I mostly want to make it faster for the "happy path" and more idiomatic (and simpler).

NoahTheDuke21:12:35

@seancorfield does your production app use java.jdbc directly? do you use any wrappers?

seancorfield21:12:10

We use java.jdbc directly. We use HoneySQL for heavy lifting in some places for composable queries (esp. reporting queries). But that just builds the sql-params vector that we pass to java.jdbc.