Fork me on GitHub
#clojure-dev
<
2019-04-30
>
alexmiller00:04:09

Generally things around interop throw rather than check

borkdude10:04:35

usually it’s better to check how you got to the npe in the first place and fix it earlier

ikitommi11:04:52

Is there still some known slowness with Java11 using Clojure 1.10.1-beta2? Run a set of micro-benchmarks and it seems to be some tens of percents slower than with java8

ikitommi11:04:22

➜  porsas git:(master) ✗ lein perf repl

nREPL server started on port 50322 on host 127.0.0.1 - 
REPL-y 0.4.3, nREPL 0.6.0
Clojure 1.10.1-beta2
OpenJDK 64-Bit Server VM 11.0.2+9

user=> (require '[porsas.core-test :as p])
user=> (p/perf-test-one)

 391 ns
 413 ns
 381 ns

ikitommi11:04:35

➜  porsas git:(master) ✗ lein perf repl

nREPL server started on port 50331 on host 127.0.0.1 - 
REPL-y 0.4.3, nREPL 0.6.0
Clojure 1.10.1-beta2
Java HotSpot(TM) 64-Bit Server VM 1.8.0_172-b11

user=> (require '[porsas.core-test :as p])
user=> (p/perf-test-one)

 335 ns
 351 ns
 347 ns

ikitommi11:04:00

(a lot of noice removed from console print)

ikitommi11:04:04

tested with other libs too, the slowdown seems consistent.

ikitommi11:04:49

on macOS 10.14.4

alexmiller11:04:35

I have not generally seen that

alexmiller11:04:19

It would be more useful to compare vs Java 11.0.1 or Clojure 1.10.0

alexmiller11:04:07

I also inherently disbelieve all timings done under lein

ikitommi13:04:58

I'll rerun with those and report if still see that.

seancorfield21:04:51

FYI, we just deployed Clojure 1.10.1 Beta 2 to production across all services.

seancorfield21:04:02

(we're still on JDK8 tho')