R devel - Apr 2004 - R INSTALL resumes on Win in case of errors (PR#6769)

I just post it not to forget - will try to contribute a fix.

Example: SparseM_0.34.tar.gz

Running
   R INSTALL SparseM
gives an error ("Execution halted"), but resumes the rest of the 
installation procedure and prints "* DONE" at the end. Hence R CMD
check
(either running solely or in maintainer's mode) cannot detect that the 
installation fails. Output:


[SNIP ...]
Creating a new generic function for "solve" in "SparseM"
[1] "solve"
Error in setMethod("model.matrix", "matrix.csc.hb",
function(object,
...) { :
     No existing definition for function "model.matrix"
Execution halted
Execution of package source for SparseM failed
mv: cannot stat `.RData': No such file or directory
   installing demos
[SNIP ...]
* DONE


Also note, that the package installs smoothly on Linux/Unix.
Looks like package "stats" is attached on Unix/Linux during the 
installation, but not so on Windows. I told the maintainer to use 
require("stats") in order to be on the save side. Nevertheless, I
think
we should try to unify the behaviour.


Uwe Ligges


--please do not edit the information below--

Version:
  platform = i386-pc-mingw32
  arch = i386
  os = mingw32
  system = i386, mingw32
  status   major = 1
  minor = 9.0
  year = 2004
  month = 04
  day = 12
  language = R

Windows NT 4.0 (build 1381) Service Pack 6

Search Path:
  .GlobalEnv, package:methods, package:stats, package:graphics, 
package:utils, Autoloads, package:base

I've fixed this.  It seems the shell used was not passing status codes in 
the same way as a real sh.exe.

Under Windows, --save installs assume only the base package.  This is
necessary to bootstrap the build, as all packages are installed with the
same code.  Under Unix they have R_DEFAULT_PACKAGES=, that is the default
set.  I think Unix is wrong here, and will change it in R-devel.

On Thu, 15 Apr 2004 ligges@statistik.uni-dortmund.de wrote:
> I just post it not to forget - will try to contribute a fix.
> 
> Example: SparseM_0.34.tar.gz
> 
> Running
>    R INSTALL SparseM
> gives an error ("Execution halted"), but resumes the rest of the 
> installation procedure and prints "* DONE" at the end. Hence R
CMD check
> (either running solely or in maintainer's mode) cannot detect that the 
> installation fails. Output:
> 
> 
> [SNIP ...]
> Creating a new generic function for "solve" in
"SparseM"
> [1] "solve"
> Error in setMethod("model.matrix", "matrix.csc.hb",
function(object,
> ...) { :
>      No existing definition for function "model.matrix"
> Execution halted
> Execution of package source for SparseM failed
> mv: cannot stat `.RData': No such file or directory
>    installing demos
> [SNIP ...]
> * DONE
> 
> 
> Also note, that the package installs smoothly on Linux/Unix.
> Looks like package "stats" is attached on Unix/Linux during the 
> installation, but not so on Windows. I told the maintainer to use 
> require("stats") in order to be on the save side. Nevertheless, I
think
> we should try to unify the behaviour.
> 
> 
> Uwe Ligges
> 
> 
> --please do not edit the information below--
> 
> Version:
>   platform = i386-pc-mingw32
>   arch = i386
>   os = mingw32
>   system = i386, mingw32
>   status >   major = 1
>   minor = 9.0
>   year = 2004
>   month = 04
>   day = 12
>   language = R
> 
> Windows NT 4.0 (build 1381) Service Pack 6
> 
> Search Path:
>   .GlobalEnv, package:methods, package:stats, package:graphics, 
> package:utils, Autoloads, package:base
> 
> ______________________________________________
> R-devel@stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-devel
> 
> 
-- 
Brian D. Ripley,                  ripley@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