Smart detection of wrap width?

A lot of this depends on what context you are running R in, e.g.,
Windows console, Mac console, or command line in a unix-alike. Or
within ESS in emacs. Those are different interfaces supported by, to
some extent, different people, and are based on the underlying
capabilities provided by the operating system.

Have you yet encountered
  options()$width
?
For example,
  options(width=100)
will cause wrapping at 100, at least for certain kinds of output.

In an xterm shell running in an X windows context, I frequently use

setwid <- function ()
{
    if (!interactive())
        return(invisible(NULL))
    scon <- pipe("stty -a")
    stty <- scan(scon, what = "", sep = ";", quiet = T)
    close(scon)
    cstr <- stty[grep("columns", stty)]
    options(width = as.numeric(gsub("[^0-9]", "", cstr, ignore.case = T)))
    paste("width =", options()$width, "\n")
}

A function I wrote that resets the width option to match the window
widths, and therefore adjusts the wrapping after I resize a windwo.

Yes, I found the width option in the help pages, but I was wondering
if there was automatic setting of the wrapping according to the
current window width.

Your function works exactly as I wished.  I'll probably get smarter
with time (I hope) but would it be reasonably good practice to stick
this into ~/.Rprofile?  I don't suppose there is a way to have it
automatically invoked when the window size/positition changes?  (It's
still priceless even without automatic triggering).

On Fri, Apr 17, 2015 at 7:20 PM, MacQueen, Don <macqueen1 at llnl.gov>
wrote:

A lot of this depends on what context you are running R in, e.g.,
Windows console, Mac console, or command line in a unix-alike. Or
within ESS in emacs. Those are different interfaces supported by, to
some extent, different people, and are based on the underlying
capabilities provided by the operating system.

Have you yet encountered
  options()$width
?
For example,
  options(width=100)
will cause wrapping at 100, at least for certain kinds of output.

In an xterm shell running in an X windows context, I frequently use

setwid <- function ()
{
    if (!interactive())
        return(invisible(NULL))
    scon <- pipe("stty -a")
    stty <- scan(scon, what = "", sep = ";", quiet = T)
    close(scon)
    cstr <- stty[grep("columns", stty)]
    options(width = as.numeric(gsub("[^0-9]", "", cstr, ignore.case = T)))
    paste("width =", options()$width, "\n")
}

A function I wrote that resets the width option to match the window
widths, and therefore adjusts the wrapping after I resize a windwo.

I don't suppose there is a way to have it
automatically invoked when the window size/positition changes?

For ESS see https://github.com/gaborcsardi/dot-emacs/blob/master/.emacs

Best,
Ista

On Apr 17, 2015 7:37 PM, "Paul Domaskis" <paul.domaskis at gmail.com> wrote:

Yes, I found the width option in the help pages, but I was wondering
if there was automatic setting of the wrapping according to the
current window width.

Your function works exactly as I wished.  I'll probably get smarter
with time (I hope) but would it be reasonably good practice to stick
this into ~/.Rprofile?  I don't suppose there is a way to have it
automatically invoked when the window size/positition changes?  (It's
still priceless even without automatic triggering).

On Fri, Apr 17, 2015 at 7:20 PM, MacQueen, Don <macqueen1 at llnl.gov>
wrote:

A lot of this depends on what context you are running R in, e.g.,
Windows console, Mac console, or command line in a unix-alike. Or
within ESS in emacs. Those are different interfaces supported by, to
some extent, different people, and are based on the underlying
capabilities provided by the operating system.

Have you yet encountered
  options()$width
?
For example,
  options(width=100)
will cause wrapping at 100, at least for certain kinds of output.

In an xterm shell running in an X windows context, I frequently use

setwid <- function ()
{
    if (!interactive())
        return(invisible(NULL))
    scon <- pipe("stty -a")
    stty <- scan(scon, what = "", sep = ";", quiet = T)
    close(scon)
    cstr <- stty[grep("columns", stty)]
    options(width = as.numeric(gsub("[^0-9]", "", cstr, ignore.case = T)))
    paste("width =", options()$width, "\n")
}

A function I wrote that resets the width option to match the window
widths, and therefore adjusts the wrapping after I resize a windwo.

Yes, I found the width option in the help pages, but I was wondering
if there was automatic setting of the wrapping according to the
current window width.

Your function works exactly as I wished.  I'll probably get smarter
with time (I hope) but would it be reasonably good practice to stick
this into ~/.Rprofile?  I don't suppose there is a way to have it
automatically invoked when the window size/positition changes?
(It's still priceless even without automatic triggering).

For ESS see
https://github.com/gaborcsardi/dot-emacs/blob/master/.emacs

Possibly, though it would take a little building.  If you were to
launch R directly when you start the xterm (loosely xterm R rather
than the default) then R would receive a SIGWINCH signal whenever
the xterm window size changes (xterm automatically sends this to its
child process).  R doesn't directly enable handling of the signal,
but there's nothing to stop you loading a dynamic library with a
little C code that set up a handler for SIGWINCH and, when it got
one, ran the equivalent of the stty command to get the new width.
The thing I've not been able to figure out is how the C code would
ever then hand that to R asynchronously.  Anyone?

At this point, and since we are in an X windows context, I think it
might be easier to use the window manager's features and write a
little macro or something that will send my setwid() command to the
active window, then assign it to a simple keystroke. Then:  resize
the window; hit the keystroke, and you're done. True, it's not fully
automatic, but it would be pretty quick and easy.

Either that or give ESS a try, using the bit that Ista offered. Or
maybe Rstudio?

Peter's got a good start, but I too would be stymied at the last
step; definitely beyond my skill.

I'm glad it's helpful!

Defining it and then invoking it in ~/.Rprofile would work, but then
you will need to be careful about managing both ./.Rprofile and
~/.Rprofile files. If you have one of the former, then the latter
does not get sourced at startup (see ?Startup). Of course, you can
put source('~/.Rprofile') in a local ./.Rprofile to take care of
that if you want.

But in the long run, it would be a better practice to put personal
helper functions like this in a package and then load it in your
.Rprofile file(s). Most of my ./.Rprofile files have

  require(rmacq)
  setwid()

in them (along with whatever other directory-specific startup
actions I want). The more personal helper functions you have, the
more valuable it will be to put them in a package instead of
defining them in ~/.Rprofile.