[Rd] lattice/points and pch="c" inconsistencies

[Deepayan Sarkar]

This is not due to grid, it's caused by code in lattice (in lplot.xy) that was 
a workaround for an old bug in grid. That bug has been fixed, but I forgot to 
change the lattice code. I'll fix this by the next release (there should be a 
patch release next week).

The actual grid behaviour can be seen by 

xyplot(y~x, 
   panel = function(x, y, ...){
      grid.points(x, y, pch = '.', default.units = "native")
      panel.abline(h = c(0, 1))
   })

Deepayan

On Thursday 12 December 2002 09:48 am, David James wrote:
> Hi,
>
> The current lattice/grid packages seem to position character
> plotting symbols (e.g., ".") differently than the base function
> points.  The following shows what I mean:
>
> ## plain old points() positions the pch="." right at y = c(0,1),
> ## and thus abline(h = c(0,1)) completely hides the points "."
> ## (provided the output device has enough resolution, like postscript)
>
> y <- rep(0:1, 50)
> x <- 1:100
>
> plot(x, y, pch = ".", cex = 2)
> abline(h = c(0, 1))
>
> ## on the other hand, lattice/grid position the "." off the
> ## the y=0,1 lines, apparently centering the character's bounding box
> ## rather than the actual glyph;
>
> xyplot(y~x,
>    panel = function(x, y, ...){
>       panel.xyplot(x, y, ...)
>       panel.abline(h = c(0, 1))
>    },
>    pch = ".", cex = 2
>    )
>
>
> I've noticed the argument "just=" in grid.text(), but its default
> value of "centre" would seem to be what's needed in this example(?)
>
> This behavior is consistent on Linux and Windows versions of R 1.6.1,
> lattice 0.6-6, and grid 0.7-2.
>
> Is this a lattice/grid feature or a bug?