Skip to main content

Here comes JDK 6u14

Posted by opinali on February 11, 2009 at 4:41 AM PST

Get it here while it's hot; check the Release Notes, pretty long one btw, 204 bugs and that's just b01. Update 14 is going to be a massive update, not in the same ballpark of 6u10, but still. (6u13 will be just an SSR.)

Because I'm a JDK-junkie, once loaded the Release Notes I noticed that this release contains a large backport of JDK 7 goodies - lots of bugtrack links were already highlighted as visited on my browser... that's good news, because it's reasonably mature stuff that Sun has been developing/testing for some good time in the JDK 7 project.

My highlights:

- The G1 ("Garbage First") collector!
- Loading of anonymous classes; described in John Rose's blog, it's one of the DaVinci Machine subprojects. Supporters of dynamic languages (and other runtimes that need intense bytecode generation) will drool on this one.
- HotSpot 14, with a sh**load of minor enhancements, fixes and new optimizations;
- Compressed pointers, a 64-bit optimization previously available in the Performance Release (so that's already production quality).
- New optimizations for BigDecimal and TreeMap/HashMap APIs.
- More enhancements apparently targeted at JavaFX, but good for Swing as well. Includes additional HW/LW fixes, as 6u12 only delivered a partial fix for this problem. BTW, even the G1 backport may be a JavaFX escalation because it's obviously good for soft-realtime work like animation.
- Minor fixes/enhancements in the new plugin, like having a single JRE setting for applets and JAWS (this complication never made any sense to me).

My favorite fix however, is for Bug 6793818: JpegImageReader is too greedy creating color profiles. Not because it's such a hugely important bug in itself, but because it represents the kind of work we need - reducing the JRE's startup time and footprint by making it smarter/lazier/loosely-coupled/componentized/whatever.

With this build (slated to ship mid-May: call it the JavaFX 1.5 Runtime Release?) I think I'll calm down my anxiety for JDK 7, as Sun is being aggressive to move forward the [now "Good Old"] JDK 6, including backports of some JDK 7 goodies that are either simple/low-risk (like the anonymous classes it seems), or mature due to an already-long development cycle (like G1).

I was only surprised to NOT see the parallel classloading fix in this build; I supposed that should come in the following "major minor update".