[self-interest] Re: Self on Intel running Solaris?

[Jecel Assumpcao Jr.]

John Ars wrote:
> What would it take to port?

Somebody with the time and the knowledge. The need is clearly there.

> Any ideas?

You could implement Self in Java so it would run everywhere (ha!) without
porting:

    http://www.ConsultAr.com/JSelf/

Or try to create a new implementation:

   http://www.linuxsupportline.com/~openself/

Or patch the Sun implementation to generate X86 code:

   http://www.cichon.de/self/

My own idea is to create a new implementation in Self (more about this in the
next email).

> What about the VM?

While there are several issues in porting the Virtual Machine, the code
generation part of the compiler (which are a part of the VM) is by far the most
complicated one.

> Is there any
> documentation out there?

There is some documentation available with the Self VM sources (download from
http://self.sunlabs.com) but mostly you have to be able to read the sources
themselves (50K lines of C++).

Note that in Self 4.0 (released in 1995) the organization of the files didn't
make it very clear what you would have to change to do a port. This is what
Gordon tried to get running on Linux X86. He got stuck in the details for stack
management, since it was so Sparc specific. For new 4.1 release, David Ungar
has made it much easier to add new platforms to the system and now you have an
example of a very non-Sparc CPU (the PowerPC) to be able to understand how
things work and what needs to be changed.

-- Jecel