[ESS-bugs] new iESS bug "Symbol's value as variable is void: resume"

[Martin Maechler]

>>>>> "VS" == Vitalie Spinu <spinuvit at gmail.com>
>>>>>     on Fri, 7 Sep 2012 17:34:58 +0200 writes:

    >>> Martin Maechler <maechler at stat.math.ethz.ch> on Fri, 7
    >>> Sep 2012 17:11:48 +0200 wrote:

    >>>>>>> "VS" == Vitalie Spinu <spinuvit at gmail.com> on Fri, 7
    >>>>>>> Sep 2012 12:15:42 +0200 writes:

    >>>>> Martin Maechler <maechler at stat.math.ethz.ch> on Fri, 7
    >>>>> Sep 2012 09:14:54 +0200 wrote:

    >>>>>>>>> "MM" == Martin Maechler
    >>>>>>>>> <maechler at stat.math.ethz.ch> on Fri, 7 Sep 2012
    >>>>>>>>> 08:42:41 +0200 writes:

    MM> Type things like

    MM> require("Rmpfr") and (a <- ......)

    MM> Then, after my typing of [Enter], *instead* of the
    MM> corresponding line being evaluated by R, it is *NOT*
    MM> evaluated, and in the minibuffer I get the error message

    MM> Symbol's value as variable is void: resume

    MM> In the *R* buffer, I get an extra newline and a new
    MM> prompt.  If I do the same thing a second time, it works.

    VS> Solved in trunk. Please keep an eye on this type of
    VS> misbehavior. It's your special environment (remote drive
    VS> + 3000 R packages installed) that might make the
    VS> difference for the delayed evaluation (now used to call
    VS> installed.packages()).

    >> aha!!

    >> I've svn up'ed and now don't get the error, but see

    >> Warning message: closing unused connection 3
    >> (/sfs/s/linux/rhel3_amd64/app/R/R_local/library_F17/fisheyeR/Meta/package.rds)
    >> in my *R* buffer after the first library() call..

    VS> ESS is just calling installed.packages(). Looks like an
    VS> R issue.

    VS> Would you see the same if you call installed.packages()
    VS> and then interrupt with C-c C-c, and call library()?


    VS> If so, then I am not sure there is an easy workaround
    VS> for this :(

Well,  don't start this "sneaky action"  immediately after
starting up R.
I kill R and R restart it often. I want this to be immediate and
as fast as possible.  That's been one of the real advantages of
ESS over other interfaces.

So, one workaround: After starting R, wait a few seconds, if the
the user does not start interacting with *R*, and only then
wait for an idle time before starting that action.
{{and it's okay if you make this configurable and let the
  default be your current behavior, which is just a problem for
  me, ...}}

Martin