[6481] in s-news-athena
Re: pairs() bug??
daemon@ATHENA.MIT.EDU (shyu@research.att.com)
Wed Feb 22 11:00:33 1995
From: shyu@research.att.com
Date: Wed, 22 Feb 95 10:14 EST
To: S-news@utstat.toronto.edu
Rich Gonzalez <gonzo@u.washington.edu> wrote :
> The pairs() code suggests that I could pass ylim & xlim arguments to the
> plot command so all the plots have the same ylim and xlim of my choosing.
> I got an error message when I tried that. That is,
>
> pairs(x,y,panel=myfunction,ylim=c(0,1), xlim=c(0,1))
>
>
> The kludge I worked out was to copy pairs() to pairs.fix(), edit the plot
> to have the specific ylim and xlim I want and it worked fine. This is not
> a general solution, of course. Anyone work out a better fix or tell me
> what I am doing wrong in passing ylim & xlim through pairs?
This is indeed a bug in the current release of S for pairs().
A while ago I posted a fix to this bug :
1. Add the following to pairs() argument list :
xlim = NULL, ylim = NULL
2. Replace the call to .S(plot()) in pairs() by the following :
.S(plot(if(!is.null(xlim)) xlim else xrange[[j]],
if(!is.null(ylim)) ylim else xrange[[i]],
type = "n", axes = F), "plot")
This may very well be the same "kludge" mentioned, but I believe it will
be general enough to solve all the xlim & ylim setting problem in pairs().
Ming
o=======================================================================o
| Ming-Jen Shyu (Ming) INTERNET: shyu@research.att.com |
| AT&T Bell Laboratories, Room 2C-263 UUCP: attmail!research!shyu |
| 600 Mountain Avenue office: (908) 582-6863 |
| Murray Hill, NJ 07974-2070, USA fax: (908) 582-3340 |
o=======================================================================o
