[Rd] inconsistent handling of NA lty (PR#3217)

[paul at stat.auckland.ac.nz]

Full_Name: 
Version: 
OS: 
Submission from: (NULL) (130.216.50.118)
Submitted by: paul


>From report by Peter Dalgaard ...

I found that if a numeric vector is used to set the line type in
segments(), then any NA values get plotted with a particular pattern,
rather than omitted as usually happens with NAs. Similarly, 

  plot(rnorm(5),type="l",lty=as.numeric(NA))

does plot a line, curiously enough not with the same pattern as

  plot(1:7,2:8)
  segments(1:5,2:6,4:8,3:7,lty=c(NA,1:4))

lty=NA (i.e. logical NA) gives an error.