smallest scheme kernel?

hi,

might anybody have pointers to small schemes? (i've just started
perusing things like Petit Larcenty and Ikarus.)

i'm seeking a scheme implementation that is as much as possible
written in scheme, with a really teeny "native" kernel. performance
isn't as important as extremely easy portability to me at the moment
e.g. to the CLR or JVM (by which i mean one could port to both, not
that the given scheme targets only one). bonus, ideally the kernel
would separate out the GC cleanly so one could plug into another
already-existing GC, like plugging in to the CLR or JVM.

thanks,
sincerely.

> Just another option to throw out there.

cool, thank you for the thoughts. i've peeked at Chez before, having
heard good things about it.

your point about it being commercial brings up a good aspect; i think
something open sourceish would probably be most preferable to me. (i'm
not super fascist anti-commercial, but at the moment for this i'm
leaning FLOSSy.)

regarding linking in to JVM/CLR, i was wondering if there'd be a small
Scheme where i could port the kernel to bytecode vs. CPU native code.
i should implement read & eval in Fan i guess ;-). ok, well, really,
i'm dreaming of not to have to build something from the ground up.

sincerely.

raould wrote:

> i'm seeking a scheme implementation that is as much as possible
> written in scheme, with a really teeny "native" kernel.

Can you expand a little bit of what you mean by this "native kernel"
and what properties of it that you're seeking?

Aziz,,,

raould schrieb:
> hi,
>
> might anybody have pointers to small schemes? (i've just started
> perusing things like Petit Larcenty and Ikarus.)
>

S9FES (R4RS) from Nils Holm is pretty small although I don't know if it
suits your needs. There are some small kernels around but most of them
afaik are R4RS or R5RS, if this maybe an issue.

http://t3x.org/bits/s9fes/index.html

S9FES has the charme that it comes with a cheap book describing the
implementation. Most kernels I know are although small vaguely
documented which may not be of much help. So maybe you should give a
little bit more details about the purpose.

Also

http://www.cs.indiana.edu/scheme-repository/imp.html

may be a good source.

HTH

-- Mario

> Can you expand a little bit of what you mean by this "native kernel"
> and what properties of it that you're seeking?

I have a possibly crazy/confused/clueless hope of having a Scheme that
is like 99% written in Scheme/macros and 1% in "native" code, so it
would be easily portable because the amount of code you have to re-
write when you port from an IBM System/390 to a Sharp Zaurus to a PS3
Cell is super small and easy.

Further, given that these days many things come with garbage
collectors, a way to use those rather than creating one on top. So you
can be more 'performant' on a given VM.

not sure if that makes any sense. :-}

Given that there are things like http://code.google.com/p/flisp/ and
others, it probably wouldn't be too hard to make one from scratch,
really. But I was hoping not to reinvent the wheel.

sincerely.

> So maybe you should give a
> little bit more details about the purpose.

sorry -- i can see i was lousy at explaining myself :-) :-) maybe the
email re: Aziz i wrote helps, maybe not.

thanks for the links from everybody, i will check into those.

sincerely.