AIX and OpenJDK
kelly.ohair at oracle.com
Tue Jan 31 15:26:40 PST 2012
On Jan 31, 2012, at 2:49 AM, Steve Poole wrote:
> hi all,
> Last year there was a discussion  about adding AIX platform support into OpenJDK. I'd like to pick up that conversation and complete this work.
> To recap the salient points of the thread:
> 1: The scale of the changes to support AIX without Hotspot are small ( ~ 300 LOC) There are an additional 19 new files that cover AIX specific build files and providing necessary support for AIX specific filesystem, virtual machine and process attributes in the same manner as is already done for Linux, Solaris and Windows etc. Generally the changes have been coded as capability based rather than platform focused. These changes are easy to understand and help towards improving platform portability.
> 2: Generally AIX is very close to both Linux and Solaris. As you would expect we will help ensure OpenJDK developers do not break things where they do not have access to an AIX machine.
> 3: This work will faciliate the porting of Hotspot to AIX but IBM intends to focus on our own JVM at this time (as you would imagine). We will make a binary of the JVM available for OpenJDK developers who want early access on AIX of ongoing work in JDK8 and AIX.
> I said I would post more when all the changes under item 2 above had been posted. That's basically now. I'd like to pick up the conversation again and resolve how to get the remaining files into OpenJDK so it's possible to build and run JDK 8 on AIX.
> I do not consider that the scale of these changes warrant a porting project all of their own. Does anyone have a reason why I shouldn't just start posting the additional files with the intention of getting them added into the main JDK8 repos?
>  http://mail.openjdk.java.net/pipermail/porters-dev/2011-October/000363.html
I'm concerned that we might be calling it a port when it seems like there is no intention of having a full port to AIX.
Even if the AIX port was just to Zero, that would at least be a port. Without a VM port, it seems like half a port.
Originally OpenJDK released with 'binary plugs' where separate binaries were provided to plug in where components
had not been open sourced or been replaced with open variants. It was painful, unpopular, and build wise unreliable.
I really don't want to see any form of that coming back to OpenJDK sources or makefiles.
(Granted we have our historic uglies left around our build processes).
David Holmes has been plying around with some 'custom' makefile logic, perhaps that could be used in some way here?
I do think that these changes warrant a separate porting project, not so much because of the # lines of changes
as the logistics of it being a completely new OS. A separate porting project could provide some benefits.
I do think that generic changes that increase the portability of the code base, without mentioning any of those
proprietary names in the ifdef's is fine for normal changes into jdk8. I'm all in favor of making the build
and source more portable. But we'd need to see that on a case by case basis.
More information about the jdk8-dev