Skip to content

Problem with starting and using R

8 messages · Duncan Murdoch, Brian Ripley, Thomas Lo

#
This reflects a problem in your locale (traditional Chinese): we cannot 
reproduce it.  Try running R in a different locale (e.g. append LC_ALL=en 
to the target when you start R).
On Sun, 7 Sep 2008, Thomas Lo wrote:

            

  
    
#
On Mon, 8 Sep 2008, Prof Brian Ripley wrote:

            
Maybe I have found this as a bug in iconv.  Please try a version of 
R-patched with svn revision 46507 or later.

  
    
#
On 08/09/2008 6:56 AM, Prof Brian Ripley wrote:
Unfortunately, something stopped the daily builds of R-patched on 
Saturday, and I haven't been in to the office yet to fix things.  So 
it's likely to be tomorrow before the Windows build of r46507 or later 
is available on CRAN.

Duncan Murdoch
#
On Mon, 8 Sep 2008, Duncan Murdoch wrote:

            
I only found r46507 this morning, as a side effect of something else.  So 
it would not be in a build until tomorrow.

  
    
1 day later
#
On Wed, 10 Sep 2008, Thomas Lo wrote:

            
Thank you for confirming this.

In the meantime I tracked down the exact problem.  The standards for a 
locale name are 'language_country.encoding', as in 'en_GB.utf8' or 
'English_United Kingdom.1252'.  In Windows XP (at least) Microsoft gives 
'Chinese_Hong Kong S.A.R..950'.  This is ambiguous, and in at least two 
places it was parsed as an encoding of 'A.R..950'.  There seem to be only 
three locales with dots in the country name (Hong Kong, Macau, U.A.E.), so 
it perhaps not surprising that this has gone undetected for so long.