[R] latex{Hmisc} cannot find xdvi on MAC OS X

Bill Northcott w.northcott at unsw.edu.au
Fri Jan 25 05:39:21 CET 2008


On 24/01/2008, at 10:00 PM, Kevin E. Thorpe wrote:
>
> When she does something like latex(describe(dataframe)), the latex
> file is produced, the dvi file is successfully written but then
> the error "sh: line 1: xdvi: command not found" appears.
>
> I started a terminal and a "which xdvi" returned the path
> where it found xdvi.  So, it appears that xdvi is in the
> PATH.

Firstly if you have questions about R on MacOS you be much more likely  
to get a useful reply on r-sig-mac rather than this list.

Secondly for the benefit of the non-Mac types here I will try to  
explain what is going on.

On most UNIXen or Linux virtually all process are launched from a  
shell process.  Hence they inherit the environment of a standard  
shell, and it makes sense to think of 'the' PATH or 'the' environment.

MacOS works differently.  Most processes are not launched from a  
shell.  The only time you really have a shell involved is if you open  
the Terminal app.  Then you get a Quartz window running a shell and  
any process you start from that shell will inherit the shell's  
environment variables.

However, if you start the R GUI app from the Finder or Dock, there is  
no shell in the process hierarchy (use '/Applications/Utilities/ 
Activity Monitor'  to view the process hierarchy) and the environment  
variables for the R app and anything it invokes are set up in  
the .plist files in the R Application bundle.  Those environment  
variables are not going to include the Tex paths like /usr/texbin or / 
usr/local/tetex/bin etc..

Three ways to get over it:
1.  Launch a command line R session from a Terminal shell.
2.  Start the GUI app from a Terminal shell with 'open /Applications/ 
R.app'
3.  Set the PATH in your R code

Finally, the way to check your PATH in R is Sys.getenv("PATH").  Which  
for the R 2.6.1 GUI app on this PowerMac G5 gives "/usr/bin:/bin:/usr/ 
sbin:/sbin:/usr/local/bin"

Hope this helps
Bill Northcott



More information about the R-help mailing list