[ESS] need help locating an old thread

Rodney Sparapani rsparapa at mcw.edu
Wed Feb 9 21:00:07 CET 2011

On 02/ 8/11 03:22 PM, Christophe Rhodes wrote:
> Well, here's the short version: by communicating using a dedicated
> protocol rather than over text (through the comint-style buffer), it is
> possible to communicate the semantics of certain outputs
> (e.g. evaluation results) rather than having to infer them after the
> fact by depending on a particular form of the prompt, or similar.
> This means that you can do fun things such as have a lattice object (or
> a linear model, or whatever) be presented in an emacs buffer as a
> picture, or a table, or whatever makes sense -- while maintaining the
> association between that buffer representation (text or image) and the
> underlying R (I almost wrote "lisp" :-) object.  So for instance you can
> use the mouse to interact with these "presentations" too, to use them as
> input to subsequent R interactions.
> Bit difficult to describe really.  The basic idea is "everything is an
> object", to contrast with the Unix approach of "everything is a stream
> of bytes".  (Emacs/comint is a bit in the middle)
> Cheers,
> Christophe

I can see that there may be a lot of pluses from such an undertaking.
But, no offense, the minuses are equally apparent.

1) an undertaking
2) the comint code is difficult to read/write, but, for the most part, 
it works and represents human-years of time on test; most bugs have
been found/fixed; new versions of comint require tweaking, but it's
just that, tweaking
3) if 1 and 2 are not enough, then I would add:  who amongst us knows


More information about the ESS-help mailing list