[self-interest] Re: Klein, ARM and Self Hardware

Jecel Assumpcao Jr. jecel at merlintec.com
Wed Dec 15 13:35:44 UTC 2010

Jan-Paul Bultmann wrote on Wed, 15 Dec 2010 20:20:23 +0100
> On Dec 15, 2010, at 7:33 PM, Adam wrote:
> > If I remember correctly, Klein doesn't have an x86 assembler or
> > compiler. It has a Sparc assembler, and a PowerPC assembler and
> > compiler. (The Apple transition to Intel happened right around the
> > time when Klein was being cancelled at Sun.)

That is what I remember too, but weren't there at least some stubs for
x86 or 68000 processors? I would have to look at it again... yeah, just
a i386 stub.

> Oh noes, that would explain why I can't get it to build... 10.6
> dropped Rosetta.

That shouldn't matter. What processor the Self VM used to support Klein
uses has not relation to what processor Klein itself understands since
you could consider Klein to be a cross compiler.

> So when we want to use it on a platform that is not only used in
> SuperComputers today we have to go the arm or 86 way. And I would
> then choose the ARM due to simplicity in the Assembler. (A Sparc
> laptop would be rad though :D).

I think at some point Self was running on the Sun SPARCstation Voyager,
which is pretty close to a laptop. And there used to be a Sparc laptop
from a company called Tadpole.

About the general issue of Self hardware, however, it is important to
remember that the major reason why the language failed to become more
popular was that it required you to buy a very specific computer to run
it on. My most expensive investment to date was exactly a Sun Ultra 5
machine in 1998 just to be able to use Self at home instead of depending
on the university. The Mac version in 2000 changed things, but this was
also a machine that few people had and by then the reputation of Self as
unavailable was hard to reverse.

Now if you are talking about gathering a group of people to restart the
development of Klein then it makes perfect sense to have all the
programmers use the same hardware as their initial target in order not
to waste time on porting issues in the begining.

Most people on this list are probably very tired of hearing about this,
but since we have newcomers I should mention that I worked on Self
specific hardware from 1998 to 2008. At that point I joined a group of
people who were already investing in Squeak Smalltalk so I changed the
focus of my designs. But I still feel that my unimplemented RISC42
processor would be a great target for Klein:


Note that at the board level, I use programmable logic (FPGAs) to
implement my SiliconSqueak processor. This means that the boards could
have RISC42 instead. Or even the Leon3, which is a very robust open
source implementation of Sparc.

-- Jecel

More information about the Self-interest mailing list