pch and mkh bug PR#225

s-harker@adfa.edu.au s-harker@adfa.edu.au
Thu, 15 Jul 1999 04:28:02 +0200 (MET DST)

I am sorry about the problem caused by my first attempt at reporting a
bug.  This is a followup to my report PR#225, hopefully being more
useful.  By not I have a better understanding of the problems being

I wish to report one bug relating to plot character and to mention two
other possible bugs (these may relate to differences between R and S)
I have noticed with R.  We have recently installed R and I am
converting some scripts I had previously used with a very old version
of S (not Splus).  The R version information is:

platform   sparc-sun-solaris2.5.1
arch       sparc                 
os         solaris2.5.1          
system     sparc, solaris2.5.1   
status.rev 0                     
major      0                     
minor      64.2                  
year       1999                  
month      July                  
day        3                     
language   R  

In the following I will be quoting some extracts from R A Becker, J M
Chambers and A R Wilks _The New S Language_, Wadsworth & Brooks/Cole
1988, isbn 0-534-09192-X.  We have a copy of this book since, although
our departmental server did not update from S to Splus, the campus
server did.  I am using this as an authority to report some of the
following as being bugs.

(1)  First the boxes drawn using `points()' via the following example

Y <- c(13.75, 10.50, 7.25, 4.0)
X <- c(1, 2, 3, 4)

are of a size that indicated that the mkh value is being ignored (the
upper set of boxes).  This can be fixed by using cex (the lower set of
boxes).  However, the S manual states that `mkh' gives the height in
inches of mark symbols drawn when `pch' is given as a number.  This
comment is also made in the help for the `par' function within R.  This
would appear to be a bug, using `cex' is correct only when giving a

(2)  The plotting of axis labels on an axis becomes confused if the
`at' vector is descending.  For exaample, consider the following:

labels <- c("A", "B", "C", "D")
Y <- c(13.75, 10.50, 7.25, 4.0)
X <- c(1, 2, 3, 4)

According to the S manual `label[i]' is plotted at coordinate `at[i]'.
Although S (or R) plot the vector in ascending order this still worked
(in old S, anyway), it may have been a feature.  However, R appears to
sort `at[i]' but not do a corresponding sort on label[i].  The R help
does not make the statement about plotting `label[i]' at coordinate
`at[i]', so this may be a difference between S and R.  It can be
overcome by appropriate pre-sorting of the two vectors.

(3)  The use of `mtext' only works for a string, not for a vector of
strings.  For example, the following commands.

labels <- c("A", "B", "C", "D")
Y <- c(13.75, 10.50, 7.25, 4.0)
X <- c(1, 2, 3, 4)
mtext(text=labels, side=4, line=0.5, at=Y)

plots only the first of the elements of `labels'.  Again, this may
reflect a difference between S and R.  The S manual states that `text'
is a character vector to be plotted and `at' is an optional vector of
positions.  However, the R help states that `text' is a character
string and `at' gives the location in user coordiates.  This can be
overcome by plotting `labels[i]' at `at[i]' within a loop.

Stephen Harker                                        s-harker@adfa.edu.au
School of Physics               Baloney Baffles brains: Eric Frank Russell
University College                http://www.adfa.edu.au/physics/s-harker/

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