R devel - Mar 2011 - R_HOME path getting munged in inst/doc/Makefile on Windows

Hello,

I have come across two separate packages that have a Makefile in inst/doc
which use the R_HOME variable.

In both cases, the path to R_HOME gets munged in such a way that commands
that include R_HOME fail on Windows:

For example, one Makefile, for the xmapcore package (
https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/xmapcore/username/password:
readonly) has this:

R=${R_HOME}/bin/R
SUITE=../cookbook/delia.R
[...]
${R} --vanilla --verbose < ${SUITE}

the output of trying to build this package includes:

* creating vignettes ... ERROR
E:\biocbld\BBS-2~1.8-B\R/bin/R --vanilla --verbose < ../cookbook/delia.R
E:biocbldBBS-2~1.8-BR/bin/R: not found
make: *** [pdf] Error 127
Error in tools::buildVignettes(dir = ".") : running 'make'
failed
Execution halted

It seems R_HOME is not getting resolved to a valid path. That's strange
because R CMD echo shows the right thing:

E:\sandbox>\biocbld\bbs-2.8-bioc\R\bin\R CMD echo %R_HOME%
e:/biocbld/bbs-2.8-bioc/R

That's a nice path with all forward slashes and no funny 8.3 paths with
tildes.  But it looks like when R_HOME is invoked in a Makefile, the
resulting path has a mix of forward and backslashes, and gets translated
into 8.3 style, and the resulting path is not valid for finding R
executables.

Note that R_HOME is defined within R; I don't also have it defined at the
shell level:

E:\sandbox>echo %R_HOME%
%R_HOME%

Any ideas?
Thanks,
Dan
> sessionInfo()
R version 2.13.0 alpha (2011-03-18 r54865)
Platform: i386-pc-mingw32/i386 (32-bit)

locale:
[1] LC_COLLATE=English_United States.1252
[2] LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

	[[alternative HTML version deleted]]

On Mar 21, 2011, at 9:07 PM, Dan Tenenbaum wrote:
> Hello,
> 
> I have come across two separate packages that have a Makefile in inst/doc
> which use the R_HOME variable.
> 
> In both cases, the path to R_HOME gets munged in such a way that commands
> that include R_HOME fail on Windows:
> 
> For example, one Makefile, for the xmapcore package (
>
https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/xmapcore/username/password:
> readonly) has this:
> 
> R=${R_HOME}/bin/R
> SUITE=../cookbook/delia.R
> [...]
> ${R} --vanilla --verbose < ${SUITE}
> 
> the output of trying to build this package includes:
> 
> * creating vignettes ... ERROR
> E:\biocbld\BBS-2~1.8-B\R/bin/R --vanilla --verbose < ../cookbook/delia.R
> E:biocbldBBS-2~1.8-BR/bin/R: not found
> make: *** [pdf] Error 127
> Error in tools::buildVignettes(dir = ".") : running
'make' failed
> Execution halted
> 
> It seems R_HOME is not getting resolved to a valid path. That's strange
> because R CMD echo shows the right thing:
> 
> E:\sandbox>\biocbld\bbs-2.8-bioc\R\bin\R CMD echo %R_HOME%
> e:/biocbld/bbs-2.8-bioc/R
> 
> That's a nice path with all forward slashes and no funny 8.3 paths with
> tildes.  But it looks like when R_HOME is invoked in a Makefile, the
> resulting path has a mix of forward and backslashes,
Nope, at least not in R from CRAN:

Makevars:
all:
	echo R_HOME: $(R_HOME)

[...]
echo R_HOME: c:/PROGRA~1/R/R-212~1.2
R_HOME: c:/PROGRA~1/R/R-212~1.2

But I see that you have custom rhome setting (BBS...) so changes are that is the
culprit - the rhome for that R build is set incorrectly to contain backslashes.

Cheers,
Simon



> and gets translated
> into 8.3 style, and the resulting path is not valid for finding R
> executables.
> 
> Note that R_HOME is defined within R; I don't also have it defined at
the
> shell level:
> 
> E:\sandbox>echo %R_HOME%
> %R_HOME%
> 
> Any ideas?
> Thanks,
> Dan
> 
>> sessionInfo()
> R version 2.13.0 alpha (2011-03-18 r54865)
> Platform: i386-pc-mingw32/i386 (32-bit)
> 
> locale:
> [1] LC_COLLATE=English_United States.1252
> [2] LC_CTYPE=English_United States.1252
> [3] LC_MONETARY=English_United States.1252
> [4] LC_NUMERIC=C
> [5] LC_TIME=English_United States.1252
> 
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
> 
> 	[[alternative HTML version deleted]]
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
> 
>

This is a bug in the package: you are required to quote paths in
Makefiles, e.g. to allow for spaces in the path.

And let's knock the issue of forward and backwards slashes on the 
head: they are equivalent on Windows.  Despite your perjorative 
comments, the supplied value of R_HOME is valid.

On Mon, 21 Mar 2011, Dan Tenenbaum wrote:
> Hello,
>
> I have come across two separate packages that have a Makefile in inst/doc
> which use the R_HOME variable.
>
> In both cases, the path to R_HOME gets munged in such a way that commands
> that include R_HOME fail on Windows:
>
> For example, one Makefile, for the xmapcore package (
>
https://hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/xmapcore/username/password:
> readonly) has this:
>
> R=${R_HOME}/bin/R
> SUITE=../cookbook/delia.R
> [...]
> ${R} --vanilla --verbose < ${SUITE}
>
> the output of trying to build this package includes:
>
> * creating vignettes ... ERROR
> E:\biocbld\BBS-2~1.8-B\R/bin/R --vanilla --verbose < ../cookbook/delia.R
> E:biocbldBBS-2~1.8-BR/bin/R: not found
> make: *** [pdf] Error 127
> Error in tools::buildVignettes(dir = ".") : running
'make' failed
> Execution halted
>
> It seems R_HOME is not getting resolved to a valid path. That's strange
> because R CMD echo shows the right thing:
>
> E:\sandbox>\biocbld\bbs-2.8-bioc\R\bin\R CMD echo %R_HOME%
> e:/biocbld/bbs-2.8-bioc/R
>
> That's a nice path with all forward slashes and no funny 8.3 paths with
> tildes.  But it looks like when R_HOME is invoked in a Makefile, the
> resulting path has a mix of forward and backslashes, and gets translated
> into 8.3 style, and the resulting path is not valid for finding R
> executables.
>
> Note that R_HOME is defined within R; I don't also have it defined at
the
> shell level:
>
> E:\sandbox>echo %R_HOME%
> %R_HOME%
>
> Any ideas?
> Thanks,
> Dan
>
>> sessionInfo()
> R version 2.13.0 alpha (2011-03-18 r54865)
> Platform: i386-pc-mingw32/i386 (32-bit)
>
> locale:
> [1] LC_COLLATE=English_United States.1252
> [2] LC_CTYPE=English_United States.1252
> [3] LC_MONETARY=English_United States.1252
> [4] LC_NUMERIC=C
> [5] LC_TIME=English_United States.1252
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
-- 
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