Comments on: String Concatenation Revisited

By: jlee

jlee — Wed, 25 Feb 2009 14:06:35 +0000

It’s the default VM. I didn’t do any tweaking (not even memory parameters). You’re right about nanoTime(), though. I always forget that method exists now. I learned the millis time years ago and stopped thinking about it. I’ll have to update that part of my playbook, though. Thanks for the reminder.

By: Mihael Vrbanec

Mihael Vrbanec — Wed, 25 Feb 2009 08:53:10 +0000

Thanks for your update and benchmarking on the topic. As you mention the JIT kicking in after two loops it would be nice to know if this is the client or server VM and if there are differences between the two.
Another side note: System.nanoTime() should be preferred to System.currentTimeMillis() for measuring short periods of time because it’s using the most precise timing available on the platform.

By: jlee

jlee — Mon, 23 Feb 2009 20:02:54 +0000

@Brian Jackson Whoops! You're right. That's a typo from a (too) quick adaptation of the early test. I'll update and rerun it...

By: Brian Jackson

Brian Jackson — Mon, 23 Feb 2009 18:33:07 +0000

Is it realistic that you’d call toString() within the loop of appendAcrossLoops()? I know it would only improve the results of something that’s already the clear winner, but I expected you to call toString() after the loop.