[Rd] Is Omegahat in R a general language function bus ?

Duncan Temple Lang duncan@research.bell-labs.com
Thu, 1 Feb 2001 07:45:15 -0500 (EST)

> Date: Tue, 30 Jan 2001 14:51:49 +0000
> From: David K Walker <dkwalker@online.no>
> Sender: owner-r-devel@stat.math.ethz.ch
> Precedence: bulk
> Hi !
> Can we e.g. embed Perl in R and embed R in Java, start the program counter
> in Java, call from Java to Perl via R, and then either (1) call back from
> Perl to Java via R or (2) just plain return from Perl to Java via R ?

Conceptually, Omegahat can act in this way as a broker or bus
connecting two other languages within R. However, in practice, I think
one would be less inclined to use this approach than finding a more
direct communication mechanism, in this case between Perl and Java.
And indeed there is such a mechanism -- JPL -- and it is more than
likeley available in your Perl installation (perhaps not compiled).
Code in these two languages could communicate with each other using
this mechanism and both could be running within R and communicating
with S code via the .Perl, .Java and callback mechanisms.

Using Omegahat as the broker would involve an extra layer of
conversion. Even if this were not an issue for performance in a
particular project, it certainly is a more complex setup to debug.  If
there were not already a tool for this pair of languages (Perl and
Java), it might indeed make sense to use Omegahat as a "quickly"
deployed prototype. 

> My reason for asking this is general interest in extensible programming
> languages.  My background is heavy in Forth, moderate in assembly language,
> so far very light in Scheme, R, Java, Perl, XML, Python.  Later I hope to
> implement new statistical software and port an existing large scale time
> series database/editor/tablewriter as packages under Debian
> Gnu/Linux.

And it is very much our goal in developing the collection of Omegahat
inter-system interface packages to facilitate this type of development
with the minimal amount of re-development. So it would be good to find
out what is needed and see how these interfaces can be adapted to such
needs.  Perhaps it would be good to discuss these on the omegahat
development list (omega-devel@omegahat.org) or directly with me.

> David Walker

 All the best,


Duncan Temple Lang                duncan@research.bell-labs.com
Bell Labs, Lucent Technologies    office: (908)582-3217
700 Mountain Avenue, Room 2C-259  fax:    (908)582-3340
Murray Hill, NJ  07974-2070       
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch