[Rd] Re: [R] early R messages to stdout

Prof Brian Ripley ripley at stats.ox.ac.uk
Tue Jun 10 12:59:58 MEST 2003


No patch was included in the distributed message.  Moved to R-devel, as 
this is not a help question.

On Sun, 8 Jun 2003, John Zedlewski wrote:

> Hi,
>  I have an R script that takes its input in the form of command-line 
> parameters. It works fine, but R complains about every unknown arg with the 
> "ARGUMENT %s ignored" message, and this goes to stdout instead of stderr 
> because R_ConsoleFile isn't set yet. Is it really necessary to process all 
> command line args before setting R_ConsoleFile? It seems that only Aqua 
> systems care about their arguments when choosing the console file.

This is not what is intended.  R_ShowMessage should be showing messages
in a suitable way: for console-based systems that should probably be an 
alert box.  By not having R_ConsoleFile set R_WriteConsole is used.
I don't think you can assume that all users of that code will have a 
stderr.

>   I've attached a diff (against 1.7.0) that fixes this issue, so that non-Aqua 
> unix folks can redirect stderr to /dev/null and not have to worry about those 
> annoying argument ignored errors anymore.

NO!  That also redirects all the R warnings and errors to /dev/null, and
is definitely undesirable.

If you find the warnings (*not* errors) annoying, don't misuse R by
supplying incorrect arguments.  Use e.g. environment variables to pass
your input.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595



More information about the R-devel mailing list