Jikes RVM

Submitted by N8Dawg on Tue, 06/08/2004 - 14:58.

I've asked this before but nobody answered so I'll ask again Smiling Is there any reason why we're not using the Jikes RVM? Would getting JNode and the Jikes RVM to play nicely be a worthy project?

‹ Jnode Progress Bringing OSGi into discussion again ›

jikes rvm installation

Submitted by apoorv_garg on Sun, 09/12/2010 - 04:45.

home/apoorv/apoorv/jikesrvm-3.1.12/build/tasks.xml:26: Compile failed; see the compiler error output for details.

not the better place

Submitted by Fabien D on Sun, 09/12/2010 - 10:46.

It's not the better place to ask : you should ask on jikesrvm web site.

Fabien

my blog : en français, in english or both

What is more portable?

Submitted by jpg on Sat, 07/31/2004 - 08:21.

I dont know much about the design of our current VM, but i think some parts of it is writen in Java (isn't true?).

Thats a strong thing of JNode VM. I suggest to keep native code as smaller as we can.

Is JikesRVM easier to port to other platforms than JNode VM? That's the criterion. Eye-wink

All java

Submitted by admin on Sat, 07/31/2004 - 11:17.

Both VM's are written fully in Java.

They both compile java bytecode on the fly to native code.
Only JikesRVM supports both Intel & PowerPC and is smarter about it's optimizations.

Ewout

Considering it

Submitted by admin on Sat, 07/31/2004 - 05:26.

I'm seriously considering replacing our VM by JikesRVM. I want to share with you the reasons for doing so and the problems I'm encountering/expecting.

First of all the reasons:
- Writing a good & fast VM is a difficult process which takes a lot of time.
- RVM has compilers for both Intel & PowerPC
- RVM has both baseline & optimizing compilers
- RVM has lots of memory management options, lots of different GC types
- A lot of universities are using it for their research, which should be fast innovation.

Now the problems:
- RVM uses an all Unix build process with bash scripts, makefiles & C tools. These should be replaced by java, ant.
- RVM is less "controllable" then our own VM due to its woner (IBM) and the user base.
- RVM is huge, getting to know it is quite difficult.

I'm currently in contact with some JikesRVM developers to see if we can overcome most of these problems.

Please note that this not mean that the decision has been made, it still is an experiment. If this experiment succeeds, we can decide if we want to switch.

Ewout

licence of jikes ?

Submitted by Fabien D on Fri, 06/11/2004 - 18:54.

What is the licence of Jikes RVM ?

I don't know if JNode VM is easily replaceable.

However I personally don't see a need to replace it except the fact that Jikes RVM is probably more stable because extensively used (I think).

Fabien

Jikes RVM and JNode

Submitted by Ian on Wed, 11/24/2004 - 15:24.

Hi,

I was wondering what's the status of things? I believe Ewout is porting MMTk, does this include porting all of Jikes RVM? I'm hoping to tidy other related projects up a bit for the release of Jikes RVM 3.0 and I hope to look at Georgios Gousios's work (his MSc is at http://www.cs.man.ac.uk/apt/projects/jamaica). I have the source to JikesNODE. Currently it integrates JNode and the JikesRVM by rewriting the JikesRVM boot image runner. He was working on integrating the different classpath versions and getting drivers to work. If I carry this on what would the plans be? Is this of interest to JNode generally? Thanks,

Ian Rogers

--
The Jamaica Project: http://www.cs.man.ac.uk/apt/projects/jamaica
The University of Manchester: http://www.manchester.ac.uk/

direct link

Submitted by Trickkiste on Wed, 11/24/2004 - 16:17.

Hi,

here is the direct link to the thesis:
JIKESNODE
I took a quick look at it and it seems to me that a major part of the exchange (JNODE VM <-> JikesRVM) allready took place. But this was achieved using a new build system (maybe it was a changed version of the Jikes RVM build system). So there are quite a few things to work on (like the inability of building JNode Plugins [they are a main concept of JNode]).

Andreas