Release 0.3 of LSE64 is now public, under the GPL. I had provided
previous releases only to Noqsi Aerospace customers. You may obtain a
tarball from http://space.mit.edu/home/jpd/ (thanks to the MIT Kavli
Institute for allowing me to continue to use their server).

I have tested this release on 32 bit PPC MacOSX and AMD64 Linux.
Previous releases also work on Pentium Linux, but I have not (yet) tried
this one on my Pentium DAQ box (it's turned off and I'm 2000 miles from
its power switch at the moment).

LSE64 is a reimplementation of Bob Goeke's 16 bit LSE, but with a 64 bit
cell size and other changes. LSE was, in turn, a stripped-down version
of Jon Sachs' STOIC. Unlike nearly all other Forths, it implements a
unified, consistent language rather than two confusingly similar
languages for interpreted commands and compiled definitions.

LSE64 a work in progress, intended for laboratory data acquisition (at
which it has proven very useful). There are rough edges, experimental
features, and other things I'm considering changing. Comments are welcome.

Very good forth. !!
Nice doc, simply instructions....

If you like I have some tricks in C for avoiding the cost of call
function in you code, simply use goto.

Pablo Reda
http://www.geocties.com/redaforth

"pablo reda" gmail.com> a

"John Doty" whispertel.LoseTheH.net> a

Charles Melice wrote:

Much of Forth's simplicity disappears if you have multiple cell sizes,
so so LSE64 doesn't. Once you have a uniform cell size, a single data
stack makes good sense,I think.

Yes. I've mentioned this here before. The idea is growing on me.

Thank you for your comments!

pablo reda wrote:

Thank you for your comments!

This whole development is application driven, and right now I'm not
finding any need for higher speed (in all applications to date, the
bottlenecks are elsewhere). But I'll keep your approach in mind.

C not has computed goto but the compiler try convert switch .. case in
this computer goto when the index are complete and sucesive and if you
put every internal definition in this function and go to from word to
word with goto you avoid the cost of funtion calling in C

See my code for an example

John Doty wrote:

We have been waiting for this release a while. Now we can see the ideas
you promote in action. I'll for sure check it out.

Amicalement,
Astrobe

"pablo reda" gmail.com> a

Jorge Acereda wrote:

I've done a rough conversion out of curiosity. The result is ~2 times
slower on ppc for the following benchmark:

t : count count * drop
tt : t iterate
10000000 tt

I don't have any webpage to upload it, so if anyone wants it email me.
I'm curious about other architectures.