porsas

Nopea kuin sika pakkasella

Spike to see how fast we can go with Clojure + JDBC. Highly Experimental.

Related dicsussion: https://clojureverse.org/t/next-jdbc-early-access/4091

Latest version

Usage

porsas provides tools for precompiling the functions to convert ResultSet into Clojure values. This enables basically Java-fast JDBC queries while using idiomatic Clojure.

(defprotocol DataMapper
  (cache [this])
  (query-one [this ^Connection connection sqlvec])
  (query [this ^Connection connection sqlvec]))

porsas.core/compile returns a porsas.core/DataMapper instance. The following options are available for the compiler:

key	description
:row	Optional function of rs->value or a [[RowCompiler]] to convert rows into values
:key	Optional function of rs-meta i->key to create key for map-results"

Note: some RowCompiler implementations (like p/rs->map) generate the code at runtime, which might not supported in all platforms like GraalVM.

Examples

A Java JDBC query

;; 630ns
(title "java")
(bench! (java-query connection "SELECT * FROM fruit"))

Compiled query functions

(def mapper (p/compile {:row (p/rs->map)}))

;; 630ns
(title "porsas: compiled & cached query")
(bench! (p/query mapper connection "SELECT * FROM fruit")))

Cached query functions

With defaults, a bit slower (non-compiled) mapper is used. Works on all platforms.

;; 1300ns
(title "porsas: cached query")
(bench! (p/query p/default-mapper connection "SELECT * FROM fruit")))

Fully Dynamic queries

(def mapper (p/compile {:cache nil)}))

;; 1500ns
(title "porsas: dynamic query")
(bench! (p/query mapper connection "SELECT * FROM fruit"))

Performance

At least an order of magnitude faster than clojure.java.jdbc, see the tests for more details.

TODO

• more tests
• batch-api
• async-api for postgresql?

License

Copyright © 2019 Metosin Oy

Distributed under the Eclipse Public License, the same as Clojure.