Re: Self for other OSes/machines -- Linux?

[cracauer@wavehh.hanse.de (Martin Cracauer)]

>    Is Self available for other OSes/machines besides Suns?  I'm wondering if
>    it's available for Linux running on x86 machines.
> 
> Sorry to disappoint, but it's only available for Sparc/{SunOS,Solaris}
> machines.  
> 
> While we have speculated about porting the system to other
> architectures, there are no definite plans as yet.  If lots of people
> said they would use it if only it was available on
> Windows/Mac/Linux/whatever, we might reconsider.

I think that Self on Linux would be a very good thing. If you consider
Self to be a system for learning good programming, it would fit what
Linux is often used for.

>From what I have seen, Linux is often chosen by young programmers who
want to learn programming and won't rely on closed packages from
Borland/Microsoft and live with the rough edges of DOS/Windows. Linux
already offers some great tools for that - standard UNIX environment
with virtual memory and networking, comfortable (but sometimes slow)
Lisp systems, Gnu Smalltalk, the ParcPlace GUI Builder and almost any
UNIX library.

I see Linux as one of the best platforms for people who's primary goal
is to use new programming techniques. SPARC and HP may have better and
more free programming tools in the moment, but because Linux is free,
people don't need to care about any school/university/company policy
when choosing what they want to learn. A Linuxable PC is easy to find.

I think Self would fit here *very* good. It is free, it is very
interesting for everyone who wants to become a better programmer by
looking on how different languages solves different problems and there
is currently no full-performance (native compiler) implementation of a
dynamic language on Linux (maybe Smalltalk/X).

The drawback of course is, that such a port would be hard to do. I
don't know about Self's internals, but many language implementations
for dynamic languages with GC have a real problem because of the lack
of registers in the 80x86 architecture. Additionally, Linux changes
rapidly and a system that relies on runtime loading and internals of
virtual memory would often break from one kernel release to the next.
On the other side, the kernel developers of Linux are very helpful and
professional. Get in touch with them and you'll realize that there
will be no lack of needed information about Linux internals.

The Linux user community is a bit less professional. A release of Self
for Linux would result in great interest, but in many questions from
people who don't know what a manual is (but how their mailer works).

IMHO, Self for Linux would mean to bring Self to a much greater user
base, but don't expect the new user base to bring a lot of useful
feedback. I vote for comp.lang.self moderated :-)

Is your primary goal to have a great user base and a well-know
language or to have a high-quality language with a high-quality
community? I like the first option.

Greetings (from a SPARC)
	Martin
-- 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Martin.Cracauer@wavehh.hanse.de, Fax. +41 40 5228536, German language accepted
 No guarantee for anything. Anyway, this posting is probably produced by one 
 of my cats stepping on the keys. No, I don't have an infinite number of cats.