[pypy-dev] JVM backend - status and future work possibilities?

Charles Oliver Nutter headius at headius.com
Thu May 12 06:39:32 CEST 2011


FWIW I would be happy to help out with advice and direction, especially if the JVM backend could be rejiggered to take advantage of the new invokedynamic bytecode in Java 7. I have been using it for JRuby and have been very pleased with it.

Anyway, I really hope the JVM backend can move forward and I am standing by to help out as I am able.

- Charlie (mobile)

On May 11, 2011, at 16:56, Dario Bertini <berdario at gmail.com> wrote:

> 2011/5/11 Michał Bendowski <michal at bendowski.pl>:
>> Welcome everyone :)
>> 
>> I was wondering what is the current status of the JVM backend. I'm
>> asking because I am considering working on it for my master's thesis -
>> I was inspired by a call to action from the "Features" page of
>> pypy.org. Unfortunately it's unclear to me how much work needs to be
>> done here. I understand that RPython translation is already
>> implemented (although might need some care, for the build process to
>> succeed more easily[1]), but the JIT backend is missing. There is also
>> no integration with JVM libraries - am I right? On the other hand
>> there already is a "reference" CLI implementation and a whole thesis
>> about it.
> 
> Hi, I'm in a similar situation here :)
> 
> I'm planning as well to work on the jvm backend for my master thesis:
> I contacted just today one of my professors who seems to be willing to
> mentor me, but I haven't started any work yet
> 
> the integration with JVM libraries basically is a prerequisite for
> starting the work on the JIT backend, I'm not sure and I don't
> actually know your university program, but here at my university for
> the master thesis we have something like 15 ects... so something like
> the integration with the JVM alone may very well be on the upper side
> of what effort we could be expect to put in the thesis
> 
> You're right about the CLI implementation, but after a huge
> refactoring inside pypy it wasn't functional anymore... Antonio Cuni
> (who is the author of the CLI backend we're talking about) fixed it,
> but I think that the ability to access CLI assemblies and class still
> has some issues...
> 
> btw, I'm very new to this project as well, and from what I gathered
> until now, Antonio seems to be the guy to ask to for questions about
> the OO type system :)
> 
> As a starting point: he suggested me to start looking into JPype and Jtool
> 
> http://jpype.sourceforge.net/
> http://blog.meresco.org/2010/03/11/integrating-java-in-python-with-jtool/
> 
> this kind of tools are needed for calling JVM code from CPython, thus
> letting us to try/debug the JVM integration when running pypy
> interpreted on top of CPython ( pypy/bin/py.py )
> 
>> 
>> I'm not familiar with pypy internals at the moment, nor am I a
>> compiler expert (yet!). I will be very grateful for any advice, as to
>> whether taking care of the JVM backend is a suitable amount of work,
>> and whether it can "scale", i.e. be divided into smaller parts that I
>> can implement along the way and stop if I'm out of time (I imagine I
>> can always find more work in PyPy).
>> 
> 
> I would also like to know if this kind of work would be feasible to
> tackle in parallel for 2 students like us :)
> 
> (btw, my presence on the irc channel has been quite lacking
> nowadays... I'll try to be more present in the next days)
> 
> byebye
> 
> :Dario
> _______________________________________________
> pypy-dev mailing list
> pypy-dev at python.org
> http://mail.python.org/mailman/listinfo/pypy-dev


More information about the pypy-dev mailing list