[self-interest] Re: ARM?

[Michael Latta]

It might be good to first define the goals (requirements) for the effort to better judge various approaches. This is where the money people need to speak up. What are they hoping to get out of this? A marketable platform? A larger user base? Innovations that can be touted or exploited? Is this research or development? Risk adverse or embracing? Without good ground rules any project will flounder. 


Michael

On Dec 10, 2011, at 12:26 PM, Mario <mario at wolczko.com> wrote:

> At the risk of further confusing the issue, another possibility would be to build a Self front end and runtime support for the Maxine VM. 
> 
> Maxine is now getting its third generation compiler, known as Graal, which so far is delivering spectacular performance given its approachability and youth. Plus we are actively pursuing the implementation of other, dynamic languages within Maxine/Graal. Assuming we succeed, the maintenance burden for Self would be reduced to just the Self-specific parts of the VM, and those would be written in Java, and could be developed in an IDE (eg Eclipse) and debugged using the Maxine inspector. I believe this is the state of the art for VM development environments.
> 
> I don't want to understate the effort ... we have not yet completed even a detailed outline of a Maxine implementation of a dynamic language ... But we are eager to go down this path and would love to have others working on complementary approaches.
> 
> To repeat what James said, there are several theses waiting to be written in this area, and much fun to had. I believe this approach could deliver a Self implementation that considerably outperforms the current one (because the compiler implements many more optimizations). 
> 
> Mario
> 
> 
> 
> 
> 
> ------------------------------------
> 
> Yahoo! Groups Links
> 
> 
>