Windows-specific: image(z,zlim=()) gives bogus colors outside range (PR#525)

bolker@zoo.ufl.edu schrieb:

  In the Windows version of R (not in the equivalent version
running on Linux RH 6.0), doing image(z,zlim=c(a,b)) with
non-trivial limits (i.e., smaller than the range of z)
plots bogus colors (randomly picked from the colormap,
perhaps?) in regions where z is outside of zlim.  (on the
Linux version those regions are just plotted in black, which
seems more appropriate)

  For example:

x <- seq(-1,1,length=50)
z1 <- expand.grid(x,x)
z2 <- matrix(apply(z1,1,function(x)sum(x^2)),nrow=50)
image(z2,zlim=c(0.5,1.5))

Additionally try

  image(z2,zlim=c(0.5,0.500001))

Gives segmentation fault on Solaris / Dr. Watson on WinNT

Regards, 
Uwe Ligges

  In the Windows version of R (not in the equivalent version
running on Linux RH 6.0), doing image(z,zlim=c(a,b)) with
non-trivial limits (i.e., smaller than the range of z)
plots bogus colors (randomly picked from the colormap,
perhaps?) in regions where z is outside of zlim.  (on the
Linux version those regions are just plotted in black, which
seems more appropriate)

  For example:

x <- seq(-1,1,length=50)
z1 <- expand.grid(x,x)
z2 <- matrix(apply(z1,1,function(x)sum(x^2)),nrow=50)
image(z2,zlim=c(0.5,1.5))

  Oops, I take it back.  I *thought* it worked on Linux, but I just tried
it and got the same problem -- and I get the segmentation fault on Linux
too.