[kaffe] Kaffe Project Status

[Dalibor Topic]

Charles L. Nelson wrote:

> Hi,
> The Apache Harmony website states that part of the Harmony development 
> team consists of Kaffe developers and with Sun's possible introduction 
> of its Java IP into the GPL, will Kaffe continue to be a relevant and 
> supported project?
>  

Hi Charles,

I believe the Kaffe developers in Harmony's team are Archie, and me. 
I've largely contributed to Harmony to get the legal framework set up 
and to get it off the ground, and have moved back to working on Kaffe since.
I believe that Archie is actively contributing in terms of JCHVM, but I 
have to admit that I've not been following the progress at Harmony very 
closely, since we don't share any code yet with it. It's a nice project, 
though,
and I think it has made great strides in the quality and scope of the 
implementation in the past year.

I don't think Harmony has an influence on Kaffe's relevance, since it 
targets a different audience (where using only code under the apache 
license is an important constraint), while at Kaffe we have a more
licensing-agnostic project (as long as it's GPL-compatible, anything 
goes, including the Apache license, once GPLv3 passes through the 
drafting process next year).
i.e. the audience intrigued by opportunities offered by Harmony would 
not have used Kaffe anyway for their projects.

Sun's JDK, otoh, will soon offer a full Java implementation under a 
license suitable for inclusion into Kaffe, so I think it will be 
interesting to try to merge in parts of Sun's implementation. Given that 
Sun has a very good JVM implementation, I would expect the relevance of 
Kaffe as a vehicle for support of programs written in the Java 
programming language for distributions on the platforms supported by 
Sun's implementations to decrease quickly, and I think that's a good 
thing. It would allow Kaffe to be developped in a more experimental 
fashion again.

My plan for the next release is to be able to run with a preinstalled 
Classpath out of the box, and have some support for its generics branch.

I'd love to see Kaffe turn into a merging shop for the transition 
towards Sun's class library implementation, and I'd love to see Kaffe's 
internals like threading being implemented via GNU Pth, the zip code 
being implemented via libzzip, using glib to do away with a lot of 
portability code, and reusing components from other VMs like Cacao's 
vmgen-generated interpreter, or the llvm-based jit from Tom Tromey's 
gcj-jit branch in the gcc tree. How much of that will be done depends on 
how much people are interested in hacking on something like it. As any 
other project, it will live on based on how well it can attract future 
contributions and develop strengths for new niches.

But as long as it's fun and useful to work on, it'll go on, no worries. :)

cheers,
dalibor topic