“We have just released Java SE 6.0 Release 1 Developer Preview 1 and it is currently only available for Intel-based Macintosh computers. We will be releasing a DP in the near future for PPC also. We just wanted to get 1.6 out as fast as possible.”
“We have just released Java SE 6.0 Release 1 Developer Preview 1 and it is currently only available for Intel-based Macintosh computers. We will be releasing a DP in the near future for PPC also. We just wanted to get 1.6 out as fast as possible.”
Apple begins treating PPC-Mac owners like second-class citizens at a rate that’s speedy even for them.
Grrrr.
Don’t worry. Most of the world still uses 1.4 to run their apps.
Which is unfortunate, because 1.5 is much, much better – for developers, but also users.
AFAIK, Apple gets the source code for the JDK from Sun and then ports it to OSX. I assume this is easier for x86, since the Sun JDK already runs on x86 hardware (solaris, linux, windows). Apple needs to do all the architecture dependent code for ppc and ppc64.
I guess that’s why they need more time. Let’s hope it does not take them as long as for the Java 5 port. But this preview release for x86 gives me good hopes for that.
I don’t beliese so, Java 6 is much about (hot spot) compiler optimization, so Apple gets the x86 compiler from sun but has to do the pcc equivalent by itself…
Just my guess, but I’m guessing that Java on Intel is already available and easier for the Apple guys to make work than a PPC version.
Wow, Java 6 already? Java 5 isn’t even in Tiger. Its a separate download.
wrong – Java 5 for Tiger has come some Weeks ago via Online Software Update-
Java 5 was released April 29, 2005 – what you got a few weeks ago was an update.
Not to be off-topic but Java 6 is incredible…if you thought 5 was a great update you will be in for a treat for 6 despite the fact it is in beta. The performance improvements are tremendous.
I concur. I’m generally not a big fan of Java applications, but I have used a Mustang snapshot on linux/x86 and I’m impressed. I think the compiler does a much better job, and the JVM also seems to feel “snappier” and lighter, but the latter may just be a feeling.
The enormous speedup I measured, is a fact. Even though the benchmarked code was not very good in my opinion: it created a lot of new objects which became garbage almost immediately. I believe the escape analysis in Mustang makes sure that this kind of objects ends up on the local stack instead of on the heap and this can make an enormous difference in some cases by taking away a whole lot of stress on the collector.
Actually, the code I am talking about, is not mine, but I looked at it and was going to optimize it manually, but then I remembered about the escape analysis in Mustang and therefore decided to try a snapshot.
Actually, stack allocation was delayed until Java 7. Java 6 will perform escape analysis and use it for lock elission, but not stack allocation, unfortunately.
Ah, ok. I read (some time ago) it was planned for 6 and assumed it was in. I wonder what caused the speedup I experienced then. I guess I should read a summary with the improvements in Java6.