Re: [lojban] Re: binary Lojban

[Ryan Keppel <keppel@pdx.edu>]

I sincerely apologize for my personal note as well as my public note. In any
case, let's get down to the technical!

I have rustled up my personal language, with the intent on making it much like
Ruby (think Perl). I'm going to put in some explicit references to lojban into
the grammar. Something beyond the internationalization concepts in Java--the
better to uniquely present Lojban.

Even some of the details of the grammar will be heavily influenced by Lojban.

The Java virtual machine or others are not really appropriate for Lojban.
Technically, you could have some subset of Lojban that could be compiled to the
JVM. However, that's really not in any way useful.

Ultimately, this will have to be done in concert with an AI as you say. I would
be interested in common machine translation data structures in looking for a
way to do this. Much of that is closed source, however.

Quoting "Mark E. Shoulson" <mark@kli.org>:

 Ryan Keppel wrote:

 > How about you match what I have given to Lojban? Huh? I am thinking that you
 > will have no reply to this. Huh? Eh?
 >
 Er, how do you know I haven't?  I've been involved in Lojban longer than
 you, since even before your first association with it.  Robin says I am
 "not far behind you," so I guess I'll have to take that as confirmation
 that I have not yet met your goal.  Although in the years I've been
 here, for all my distractions and everything, I think I have contributed
 a whole lot of intellectual energy and resources to The Cause, maybe
 even more than you have.

 At any rate, all that is neither here nor there; it doesn't matter the
 tiniest bit even if I had contributed nothing up till now.  It doesn't
 speak to what you were talking about.

 To the topic at hand, though: you were asking about a "binary" version
 of Lojban, in the same sense that Python bytecode is a binary version of
 Python.  It doesn't even really matter what the bytecode is, though,
 since I suppose one could write any number of languages not at all like
 Python that still compile to Python bytecode, or to Java bytecode, and
 so on.  I suppose we could use one of those for Lojban's bytecode as
 well, though it would be a very restricted form of the language (since
 most of the gismu of Lojban don't have much meaning in terms of computer
 operations).  A higher-level Lojbanic bytecode for interpretation by and
 as an AI would be nice, but I think we can't do anything meaningful in
 that direction until first we have some decent AI working with and from
 Lojban as it is.

 We could try to write a Lojban-to-JVM compiler, I suppose, essentially
 making Lojban a programming language.  Or write a Lojban-oriented VM,
 though I don't really see what such a thing would be.  I think someone
 once wrote a Lojban-to-Prolog compiler or interpreter, which is sort of
 along those lines.

 I still somehow doubt any of these solutions are what you are looking
 for, which makes me think I'm not understanding what you want very well.

 ~mark



 To unsubscribe from this list, send mail to lojban-list-request@lojban.org
 with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
 you're really stuck, send mail to secretary@lojban.org for help.