Hi Alex,

> Which version (32 or 64 bits) do you intend to use?

I have 32 bit Ubuntu installed on 64 bit hardware. I guess I should change
to 64 bit linux, since the 64 bit version of PicoLisp seems to be really
good for interacting with foreign libraries. But thats a complete new
installation, and I first have to finish one project before I attempt that.
Until the middle of May I´m on the road without access to my own machine or
a running Picolisp, but I already try to figure out the magic behind those
extremely simple looking files like lib/math64.l or Jon´s 64-bit

My goal is to contribute two (64 bit) libraries, lib/r.l and lib/grass.l,
that enable direct access of R and GRASS GIS from PicoLisp, similiar to the
openGL library. That would make PicoLisp a fantastic environment for
business applications with special needs for Statistics and GIS. Today,
Python is more or less the `official´ scripting language for both, R and
GRASS GIS, and R is much used to analyse the spatial-temporal data from
GRASS. So R and GRASS are somehow natural allies.

BTW, there is an active discussion about a new Lisp-based version of R (
http://www.stat.auckland.ac.nz/~ihaka/downloads/Compstat-2008.pdf) by the
R-founders themselves, that should be better suited for general application
programming and for the management of huge datasets (and should be faster in
processing huge amounts of data).

> > "R can be built as a shared library1 if configured with
> ‘--enable-R-shlib’.
> > This shared
> > library can be used to run R from alternative front-end programs."
> This looks good. You should be able to call that library directly from
> the 64-bit version, analog to "lib/openGl.l" (or "lib/math64.l" for
> another example).
It is that analogy that I still don´t understand, since these two example
files only have some constant definitions and a lot of Picolisp function
definitions that do nothing else but use their arguments to call a native
function with the same number of arguments.

But there are a lot of special geographic data types in a GIS, and R has
more primitive types than just lists, and it has classes that specify
special objects that can be used as function parameters and return values.
I would expect the first step for building an PicoLisp-R or Picolisp-GRASS
interface would be to map all possible data types in that languages to
PicoLisp classes, and then write the calls to the native functions.
Therefore, the openGL file looks a bit magic to me - do the native functions
return nothing, and do they only use numbers or lists as parameters, no need
for special data-types?

I´m very much a newbie with (Pico)Lisp, still learning the basics, and I
don´t want to announce contributions here that I can´t deliver later on, but
it looks doable and I´m quite enthusiastic about a possible integration of R
and GRASS with such a great application building tool like PicoLisp.

> As my daughter must use R in her studies in the near future, and I
> learned in the past that my daughters tend to ask me for help on
> computer-related problems, it might be good if I also engage a bit on
> that subject ;-)

That would be a great coincidence ... ;)


> UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe

Reply via email to