[ESS] need help locating an old thread

[Rodney Sparapani]

On 02/10/11 10:19 AM, A.J. Rossini wrote:
>
> I'll take a stab at explaining it better -- basically, the client/R
> side ("swankR") could do something like expand the output of
>
> lm(y~x)
>
> into:
>
> the object to be returned
>
> audio to be routed via emacspeak to the speaker (voice or appropriate music)
>
> visuals to be inserted into appropriate buffers
>
> summaries to be split into a different buffer
>
> or some other cleanly parseable protocol, and know that the protocol
> is "proper".    This also helps with automagic debugging (parsing R
> exceptions and errors into appropriate places, where they can be
> computed on and Emacs probes R and the source for more info, etc).
>
> Basically, it makes computations and automagic responses easier to setup.
>
> But probably I've been opaque again.
>
> best,
> -tony

Yes, but I think I see now.  It would allow you to pass objects back and 
forth intelligently such as data frames, graphics, packages, etc.  Sort 
of a CORBA translation layer between elisp and R.  If it could
be written more simply than ESS[R], then it might have a chance.  I bet 
it would make things like R2WinBUGS much easier.

Since we are on the subject, I wonder what it would take to compile R 
into emacs directly; sort of like what has been done in BRUGS and/or 
TeXmacs.  Then you would have a rich grammar of both R and elisp 
functions at your disposal (or R based on elisp primitives I suppose). 
SXEmacs has the ability to do stuff like that, but I have never used it 
since it is Linux only.  Maybe I'll give it a try as we migrate to Linux.

Rodney