"Converge [the] best features of HotSpot and JRockit" - what does this mean for OpenJDK?

[Volker Simonis]

On Fri, Apr 16, 2010 at 1:04 PM, Andrew John Hughes <ahughes at redhat.com> wrote:
> On 1 February 2010 17:41, David Herron <david at davidherron.com> wrote:
>> In Theory ...
>>
>> Theoretically JRockit could be changed to implement the same interfaces
>> HotSpot implements so that JRocket could plug into the OpenJDK runtime.
>>  That would allow any VM to plug into the OpenJDK runtime.  This was the
>> goal of the Clean VM Interface project (
>> http://openjdk.java.net/projects/cvmi/)
>>
>> Right?
>>
>
> Right.  In fact, I would guess it doesn't need that much of a change
> as JRockit presumably uses Sun's class library already.  I doubt they
> implemented their own; those of us who worked on GNU Classpath &
> friends know how much work this is and what a never-ending game of
> catch-up it is.
>
> I don't follow Volker's sentiments about OpenJDK 'degenerat[ing] to a
> class library'.  OpenJDK is already split into a class library and VM.
>  Only the HotSpot repo. contains the VM while the other repos are all
> related to the class library.  They are also worked on by a largely
> orthogonal set of people, which is something that tends to happen
> generally; VM work attracts different mindsets to class library work.
> The less ties there are between HotSpot and the OpenJDK class library,
> the more input we will get from different VM projects and this can
> only increase the portability and flexibility of the class library.

But if you look at it realistically, there have been NO*** substantial
external contributions to the HotSpot other than perhaps a few small
bug fixes or "one-liner" improvements. And I also can't remember any
involvement of other VM projects (other than having a beer with their
authors at FOSDEM, which shouldn't be underestimate though:)

(***) sorry Gary, there's of course your Zero extension. Altough I
like and appreciate it very much, I don't think it is really
competitive with "native" VM-implementations on IA64, PPC or s390.
Please blame me if I'm wrong!

> I think Volker is confusing Classpath's support for multiple VMs
> (which I regard as an advantage over OpenJDK) with the fact the
> Classpath lags so far behind OpenJDK (which is an issue both of too
> few people working on the project and of having to wait until a new
> standard is finalised before work can be done to implement it).
> That's always an issue with a reference implementation compared to
> other implementations, and OpenJDK6 equally lags behind the
> proprietary JDK6 library (with an even smaller team than Classpath
> had).  I hope with 7 they will run more in tandem.
>

All my comments were related to the VM part. I do not follow the
other, Java library/runtime related mailing lists. Nevertheless I
think that the library/runtime part could be done more easily by the
"community".

> To go back to the idea of JRockIt and HotSpot merging, it pays to be
> wary of management speak.  There's no reason to make the assumption
> that this even means merging on a technical level.  Having both sets
> of people working together (which we've not yet seen) and sharing
> ideas would bring benefits without having to fork whole codebases.
> Let's see what happens.
>

I'm just afraid the merging will take place on a "financial" level,
i.e. there will be no budget for two VM teams in the near future. So
the question is if there will be more or less full-time Hotspot
developers in the end. I don't want to be overly pessimistic here, but
especially in the VM area I currently don't see a community which
could step in here if it will turn out that there will be less people
from Oracle working on HotSpot. I strongly believe that such a project
can not be run successfully without considerable financial funding...

>> - David Herron
>> http://davidherron.com
>>
>> On Mon, Feb 1, 2010 at 2:43 AM, Volker Simonis <volker.simonis at gmail.com>wrote:
>>
>>> In its "Oracle + Sun: Transforming the IT Industry" strategy outlook
>>> Executive Vice President Thomas Kurian of Oracle mentioned that Oracle
>>> plans to "Converge [the] best features of HotSpot and JRockit" (slide
>>> 10 of
>>> http://www.oracle.com/ocom/groups/public/@ocom/documents/webcontent/044523.pdf
>>> ).
>>>
>>> Has anybody thought about what implications this could have for OpenJDK?
>>>
>>> My fears are that if HotSpot and JRockit will "converge" this means
>>> that OpenJDK will be abandoned, because JRockit isn't OpenSource and I
>>> haven't heared of any plans to open source it any time soon.
>>> Consequently, if JRockit is and will stay closed source, a convergence
>>> of HotSpot and JRockit can only happen in a "closed" version.
>>>
>>> Are there any sentiments regarding this topic?
>>>
>>> Regards,
>>> Volker
>>>
>>
>
>
>
> --
> Andrew :-)
>
> Free Java Software Engineer
> Red Hat, Inc. (http://www.redhat.com)
>
> Support Free Java!
> Contribute to GNU Classpath and the OpenJDK
> http://www.gnu.org/software/classpath
> http://openjdk.java.net
>
> PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
> Fingerprint: F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8
>