[R] recovering "status" after q(status=n) in Windows

[Prof Brian Ripley]

On Mon, 26 Nov 2007, Dennis Fisher wrote:

> Colleagues,
>
> I am writing scripts that run on both Windows (Vista) and Linux
> (RedHat 9).  The scripts will be used in both S-Plus (version 8) and
> R (2.6.0).  I am trying to use the "status" portion of the quit command:
> ------------------------------------------------------------------------
> -----------------------------------------------------------------------
>> quit                  package:base                  R Documentation
>>
>> Terminate an R Session
>>
>> Description:
>>
>>      The function 'quit' or its alias 'q' terminate the current R
>>      session.
>>
>> Usage:
>>
>>      quit(save = "default", status = 0, runLast = TRUE)
>>         q(save = "default", status = 0, runLast = TRUE)
>>      .Last <- function(x) { ...... }
>>
>> Arguments:
>>
>>     save: a character string indicating whether the environment
>>           (workspace) should be saved, one of '"no"', '"yes"', '"ask"'
>>           or '"default"'.
>>
>>   status: the (numerical) error status to be returned to the operating
>>           system, where relevant. Conventionally '0' indicates
>>           successful completion.
>>
>>
>>
>>
>>      Some error statuses are used by R itself.  The default error
>>      handler for non-interactive use effectively calls 'q("no", 1,
>>      FALSE)' and returns error code 1.  Error status 2 is used for R
>>      'suicide', that is a catastrophic failure, and other small
>> numbers
>>      are used by specific ports for initialization failures.  It is
>>      recommended that users choose statuses of 10 or more.
>>
>>      Valid values of 'status' are system-dependent, but '0:255' are
>>      normally valid.
> ------------------------------------------------------------------------
> -----------------------------------------------------------------------
> Of the four combinations (R/S-Plus crossed with Windows/Linux), the
> only combination that I cannot get to work properly is R/Windows.
>
> The code below from Linux/R:
>>>  q(status=34, "no")
>> P2: 72> echo $status
>> 34
> shows the expected response.
>
> When I replicate that R code in Windows, then type:
> 1.  echo %status%: 		Windows echoes %status, indicating that the
> environment variable status was never set
> 2.  echo %error level%: 	Windows echoes 1 (if i execute a normal quit
> command, i.e., without setting status, Windows echoes 0).
>
> So, it appears that in Windows, %errorlevel% is the (undocumented)
> equivalent of $status (Linux).

Actually, $status is shell-specific: $? is more standard (it comes from 
sh, $status from tcsh).  $status does not work in /bin/sh on Linux for me 
(and I don't think it should).

And that %errorlevel% does contain the status on some Windows shells is 
documented in the shell documentation.

> However, in Windows, the value
> returned appears to be limited to either 0 (normal exit) or 1 (status
> set to a non-zero value) despite the claim that "valid values ...
> [are] 0:255 (as is the case in Linux).

The claim is of course correct!  E.g. in tcsh

> q("no", status=34)
[d:/R/svn/trunk/src/gnuwin32]% echo $status
34

and in cmd.exe

> q("no", status=34)

D:\R\R-2.6.1\bin>echo %errorlevel%
34

Note those examples are from Rterm: Rgui is a Windows GUI application and 
the shell seeems to receive an error status from the GUI interface layer.
But then Rterm is intended for use in scripts, not Rgui.

-- 
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