R devel - Nov 2002 - Wandering usr values in par(no.readonly=TRUE) (PR#2283)

>>>>> "Marc" == Marc Schwartz
<mschwartz@medanalytics.com>
>>>>>     on Wed, 13 Nov 2002 11:01:49 -0600 writes:
    Marc> Marc Schwartz wrote:
    >>  SNIP
    >> 
    >> I am guessing that this may be the result of par("ylog")
    >> and par("xlog") being read only and perhaps impacting the
    >> restoration of the par("usr") values when the plotting
    >> device is still open, but that may be incorrect.

    Marc> In follow up to my own note from last evening on this
    Marc> and now that I have had a couple of cups of coffee
    Marc> this morning, I need to partially correct my above
    Marc> statement and point to a presumed typo in the ?par
    Marc> help file.  For confirmation, this is R 1.6.1 under
    Marc> WinXP Pro.

    Marc> par("xlog") and par("ylog") are not Read Only
    Marc> parameters as listed in the help file and can be set
    Marc> by the user.  Indeed this is confirmed in the comments
    Marc> in par.r:

    Marc> #.Pars.readonly <-
c("cin","cra","csi","cxy","din")

    Marc> I noted this today as I was comparing the output of
    Marc> par() with par(no.readonly = TRUE) and noted that the
    Marc> latter returns '$xlog' and $ylog' as being
    Marc> settable. The other 5 par values listed in ?par as
    Marc> "R.O." (cin, cra, csi, cxy and din) return errors when
    Marc> one attempts to set them with par(args), however xlog
    Marc> and ylog do not.

Yes, indeed, and I -- as an R-core member -- am astonished:.
I didn't know it really worked, but it does, e.g., 

  > plot(1:10); par("usr")
  [1]  0.64 10.36  0.64 10.36
  > par(xlog = TRUE); par(c("xlog","usr"))
  $xlog
  [1] TRUE

  $usr
  [1] -0.1938200  1.0153598  0.6400000 10.3600000

  > points(10^(1:9), 1 + 1:9, col = "red") # draws red points above


I'm now re-directing this back to R-bugs to make sure that 
the only (right?) remaining bug, documentation in par.Rd, 
will be fixed -- and hope to close this as a bug report.

    Marc>   ..........
    Marc>   <more good analysis>

Thank you Marc!

Martin Maechler <maechler@stat.math.ethz.ch>
http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Martin wrote:
> SNIP
> 
> Thank you Marc!
Glad to be of help Martin.  

For the sake of completeness, I ran the same examples on my Linux (RH 8)
install of R 1.6.1 and observed the same behavior.

Best regards,

Marc Schwartz


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

Martin wrote:
> I'm now re-directing this back to R-bugs to make sure that
> the only (right?) remaining bug, documentation in par.Rd,
> will be fixed -- and hope to close this as a bug report.
Martin,

My apologies, as I may have missed this in my first reading of your
reply, but I wanted to be sure that it was clear that there are two
separate bugs here. The first, which I found this morning in the process
of tracking down Jari's issue and the second which Jari observed in his
initial post, which started this thread yesterday.

Bug 1. The documentation in par.Rd needs to be corrected in that
par("xlog") and par("ylog") are not R.O. I presume that this
is what you
are referring to above.


Bug 2. The improper restoration of par("usr") from saved values after
using log scaled axes. This is a result of the sequence of the code in
the par.c Specify() function (ie. par("usr") being set before
par("xlog") and par("ylog") are being set back to FALSE).

Possible solutions:

A. The code in par.c needs to be changed (possibly as I suggested
earlier) to enable the proper restoration of par("usr") after log
scaled
axes are used

OR

B. If this is not possible due to causing other problems, par.Rd needs
to have a warning note added to inform users that par("usr") will not
be
properly restored after log scales are used and that a possible solution
would be to explicitly set par(xlog = FALSE) and par(ylog = FALSE)
before restoring 'par' from saved values.  This solution of course falls
short if a function using this approach exits prematurely. One could
include code in the on.exit() function to reset 'xlog' and
'ylog' before
restoring 'par'. For example, using Jari's initial function
plot2.log():

plot2.log <- function(x=1:2,y=1:2) 
{
  op <- par(no.readonly = TRUE)

  exit.restore <- function()
  {
    par(xlog =FALSE)
    par(ylog = FALSE)
    par(op)
  }

  on.exit(exit.restore())

  plot(x, y, log = "y")
}


Again, my apologies if I am mis-reading the section from your reply
above.

Regards,

Marc Schwartz



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To:
r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._