Skip to content

matplot(... panel.first=..) fails <==> dealing with "..." (PR#267)

2 messages · Martin Maechler, Peter Dalgaard

Original
Martin Maechler
# 
Maybe, the following is not a bug in the strict sense,....

The following code shows what the effect :

  p1 <- function(...) plot.default(...)
  p2 <- function(...) { n <- names(list(...)); plot.default(...) }

  par(mfrow=c(2,2))
  p1(1:10, panel.first = grid(10,10)) # works okay (grid drawn)
  p2(1:10, panel.first = grid(10,10)) # doesn't draw the grid

A consequence of this is that the panel.first argument (of plot.default)
doesn't work with matplot()
	[well, matplot()'s doc doesn't claim it should, but still

-------
The real questions behind are:

1a) Why is "..." changed at the moment that I do  list(...)
1b) is this basic behavior desirable / changeable ....

2) How can one work around? 
   {save "..." to something else, first? 
    or explicitly save panel.first?}

-----------------------------

R versions,  both 0.64.2 and 0.65.0  (probably all R versions)

Version:
 platform = sparc-sun-solaris2.5
 arch = sparc
 os = solaris2.5
 system = sparc, solaris2.5
 status =
 status.rev = 0
 major = 0
 minor = 64.2
 year = 1999
 month = July
 day = 3
 language = R

Search Path:
 .GlobalEnv, Autoloads, package:base

--
Martin Maechler <maechler@stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
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
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Peter Dalgaard
# 
maechler@stat.math.ethz.ch writes:
...
That's a result of lazy evaluation. plot.default is really playing
with fire, the way it is currently written. list(...) will force
evaluation of the promises which in turn causes grid() to be called
too early. I don't think it is desirable to change this behavior
(well, Robert wants us to consider eager evaluation, but that would
cause the construction to stop working in all cases!)
Neither of the above, because the object is to access "..." without forcing
evaluation of its contents. One can do things like
function (...) 
{
    n <- names(match.call(expand.dots = F)[["..."]])
    plot.default(...)
}

Or, one can modify plot.default to take panel.first as an
*unevaluated* expression (mode "call" or "expression"), and do an
explicit eval(panel.first). But then you'd have to do 

p2(1:10, panel.first = quote(grid(10,10))) # or expression( )