[ESS] Feature request -- start R automagically

[Martin Maechler]

>>>>> "MM" == Martin Maechler <maechler at stat.math.ethz.ch>
>>>>>     on Sat, 24 May 2008 10:53:09 +0200 writes:

    MM> Salut Vincent, (et al)
>>>>> "VG" == Vincent Goulet <vincent.goulet at act.ulaval.ca>
>>>>> on Fri, 23 May 2008 09:03:41 -0400 writes:

    VG> Hi Martin, Le ven. 23 mai à 02:45, Martin Maechler a
    VG> écrit :

    >>>>>>>> "MM" == Martin Maechler
    >>>>>>>> <maechler at stat.math.ethz.ch> on Fri, 9 May 2008
    >>>>>>>> 17:35:56 +0200 writes:
    >>> 
    >>>>>>>> "MR" == Mathieu Ribatet <mathieu.ribatet at epfl.ch>
    >>>>>>>> on Thu, 08 May 2008 17:37:32 +0200 writes:
    >>> 
    MR> Dear list, Will it be possible to add the following
    MR> request.  When running emacs with a R script file (or
    MR> any other source
    >>> file) and
    MR> trying to C-c C-r a region while no R process is
    MR> running, you
    >>> get the
    MR> following message "Process R is not running".
    >>> 
    MR> I think it may be convenient if then ESS propose you to
    >>> launch a new R
    MR> process.
    >>> 
    MM> I agree, and I had added this feature already a few
    MM> months ago in a different context (Sweave editing) on
    MM> 2007-05-03 { according to lisp/Changelog }.
    >>> 
    MM> I hope to find time to hack this during the prolonged
    >>> Pentecoste
    MM> weekend.
    >>> 
    >>> and I did not have time then, but last night.
    >>> 
    >>> I've commited a change to ESS' development version which
    >>> automatically starts R if - you are editing an S
    >>> language buffer [ ess-language is "S" ] which should be
    >>> the case for all R source files - if your buffer has no
    >>> associated ess process.

    VG> Shouldn't it be "if there is no *running* ess process" ?

    MM> well, that's yet another feature request !  :-) which
    MM> makes sense, but is not yet implemented.

    MM> I'll see if I can implement it "quickly".  Martin

It was not quickly,
and I don't quite like the result yet,
but seem not to be able to get it working the way I want 
(at least not without resorting to horrible hacks).

The current (after you get ESS [ r3905 ]
    	     from https://svn.r-project.org/ESS/trunk/ )
problem is the following:

Assume an emacs session (in 1 frame) with 2 buffers:
Top buffer   : foobar.R
Bottom buffer: *R*  where you have executed C-c C-q  or  q("no")

Now in foobar.R, send a line (C-c C-n) to *R*
this now restarts R quite nicely,
*BUT* I end up with 
- the two buffers are reversed : 
  *R* is now top,  foobar.R is bottom
- Point ("the cursor" / focus) is in the *R* buffer instead of
  in foobar.R at the location where I had sent my line.

This happens inspite of (save-excursion ..) used in many places.
If any good Emacs lisp programmers see a solution...

Martin

    >>> 
    >>> Consequently, this feature will be in ESS 5.3.8
    >>> 
    >>> Martin Maechler, ETH Zurich