Hi,

I wrote a small but handy framework in C for
bootstrapping a Forth system, thus for
creating a "Forth Kernel". Hence the name,
even when it's already taken by a Forth
boot-loader for BSD systems(when you know a
better name let me know it).

How do you create a "Forth Kernel", thus a
small forth system for building the rest of the
system? You write your custom primitives in C,
add a bit of your own startup code to it,
compile it and then you have the "Forth Kernel".

Why? Because I wanted to start my own
forth system through writing in itself, without
ANS Forths, etc. flavor in it, just my own
ideas.

Whoever is interested into it:
http://gitorious.org/projects/bootforth

I'm interested in your opinions or ideas about it.

jpb wrote:

Hi, I've just built the sample on Windows and it works! I already had
mingw and git installed.

I totally understand your idea. I will look into the source now.

Do you think it's possible to make BootForth run on an embedded system
with 1.5K of RAM and 32K of ROM?

Best wishes,
Nickolai Leschov

Nickolai Leschov gmail.com> wrote:

Good to know. So it seems quite portable.

Hopefully you understand it. Because most people who I'm talking to
about it, don't get the point of it at the first explanation try.

Well, I should take a better name than "BootForth", even when it means
in that case bootstrapping a Forth System and not booting a computer.

At the current moment, I don't think so. It depends on a couple of
standard C functions which need something like a library.
But why, should you want to run BootForth on an
embedded system with those limitations?

For the joy? Or because you can? Both are good targets
and probably it would fit into the ROM, but the RAM
is too small and it assumes the whole memory space
for it's own, which means that it's not ROM-able.

On Jul 30, 9:19 am, jpb wrote:

Maybe the fact that "BootXYZ" is normally read as meaning booting the
computer is the problem.

Forthstrap?

jpb wrote:

Maybe call it BootstrapForth?

jpb wrote:

Embedded processors (microcontrollers) nowadays are normally programmed
in C. (sometimes, C++ even) Therefore, there is standard library. I hope
that there would be just a few things to consider to be able to run
Forth on a microcontroller; here's the ones that come to mind:
1. Using malloc/free is impractical, even though it may be implemented
in standard library.
2. IO device is a serial line, which may be used by the application as
well. Again, this may or may not be accounted for in the implementation
of the standard library.

I think Forth could fit such systems. I would like to be able to use
umbilical Forth on such system. Currently I'm stuck with C's
edit/compile/test cycle. In-curcuit debugging is not available and
programming (writing binary code) is not instantaneous and is apart from ...

"Nickolai Leschov" gmail.com> wrote in message
news:g6s397$kac$1@aioe.org...

Nickolai Leschov wrote:

Assuming you have flash instead of real ROM, it's quite possible in
SwiftX (which doesn't use any C).

...

You should really check out one of the evaluation versions of SwiftX and
see how its umbilical system works.

On Jul 31, 1:30 pm, Elizabeth D Rather forth.com> wrote:

On Wed, 30 Jul 2008 13:39:01 +0400, Nickolai Leschov
gmail.com> wrote:

An MPE cross compiler can build a full interactive Forth in
that space! There's an MSP430 somewhere on my desk ...

Stephen