I was working on an issue (enhancement) today in my groan R-package today that required adding additional plotting elements via lines()
and points()
to a device that had already been partitioned by layout()
. The code I wanted to use was essentially:
# Y and S are lists of xy.coords() objects of the same length
lyt = matrix(1:length(Y), ncol=10)
layout(mat=lyt)
# function A
# plot Y first as points
lapply(Y, function(x) {
plot(x, ...)
})
# function B
# overlay S as lines on the grid of plots for Y
lapply(S, function(x){
lines(x, ...)
})
However, I would only get all of the above lines in one subplot. For a brief moment, I considered rewriting my whole set of plotting methods to use split.screen()
or par(mfcol)
. Ugh!
On a whim, I decided to check what par('mfg')
would return after a device had been partitioned and plotted to with:
layout(matrix(1:9, nrow=3))
par(mar=c(0,0,0,0))
plot(runif(10))
I was pleasantly surprised to find:
> par('mfg')
[1] 1 1 3 3
indicating that I could potentially direct the next plot in a layout()
'ed device by setting the value of mfg=
to the next plot id:
lyt = matrix(1:9, nrow=3)
par(mfg=which(lyt == NextPlotID, arr.ind=TRUE)[1,])
Unicorns and rainbows, this works! (despite all the dire warnings in the documentation regarding incompatibilities)
Thus, the resulting code:
# Y and S are lists of xy.coords() objects of the same length
lyt = matrix(1:length(Y), ncol=10)
layout(mat=lyt)
# function A
# plot Y first as points
lapply(Y, function(x) {
plot(x, ...)
})
# function B
# overlay S as lines on the grid of plots for Y
lapply(seq_along(S), function(n){
par(mfg=which(lyt == n, arr.ind=TRUE)[1,]) # sets next plot in grid!
lines(S[[n]], ...)
})
and issue resolved.
Written with StackEdit.
No comments:
Post a Comment