Questions / Suggestions about the benchmarks using jOOQ

[lukaseder]

I have a few questions / suggestions about the benchmarks using jOOQ:
https://github.com/TechEmpower/FrameworkBenchmarks/tree/master/frameworks/Java/play2-java/play2-java-jooq-hikaricp

• Why is it using a 4 year old jOOQ version? (see https://github.com/TechEmpower/FrameworkBenchmarks/blob/master/frameworks/Java/play2-java/play2-java-jooq-hikaricp/build.sbt) There are frequent improvements to the jOOQ code, also in terms of performance. I'd be curious to see results for a much more recent version
• The query producing FortuneRecord in Application.fortunes() is using the API very un-fortune-ately, so to speak: https://github.com/TechEmpower/FrameworkBenchmarks/blob/547370f/frameworks/Java/play2-java/play2-java-jooq-hikaricp/app/controllers/Application.java#L55
  • It is very unusual to re-create a DSLContext instance per query execution. Especially when running in a benchmark, I would expect the DSLContext to be injected or otherwise cached. Its underlying Configuration contains a reflection cache, that can't be used in this particular case when calling fetchInto(FortuneRecord.class) (this applies to all methods!)
  • All FORTUNE columns are being projected in select(), so why not just use selectFrom(FORTUNE) to get a FortuneRecord without the unnecessary reflection?

[NateBrady23]

@lukaseder You're welcome to update this implementation with a PR. We don't maintain the implementations, we just review that they fall within the test rules and benchmark them. For questions about specific implementations, you'd have to reach out to the person who made the commits.

[lukaseder]

Thanks for the feedback. I will (PR), and I did (reach out).

[lukaseder]

I will (PR)

Correction: Well, I can't say I can get anything to run out of the box on my windows machine, and I'm not convinced that spending the time is worth it.