[Rd] r-devel tarball build failure on windows

[Prof Brian Ripley]

I realized that there was a different point tacked on here that has 
not been answered:

On Tue, 10 Mar 2009, Hiroyuki Kawakatsu wrote:

[...]

> By the way, I noticed that -make all recommended- still builds the CHM
> help files for certain packages (e.g. Matrix) even though I set
> USE_CHM=FALSE in MkRules. Is this expected?

Yes, it was expected.  MkRules is only used for building R, not 
installing packages.  The default for installing contributed packages 
is to build CHM help if and only if the Help Compiler is available, so 
all recommended packages followed that rule.

We haven't made a final decision yet whether to support CHM help in 
future versions (and one point R-devel did not do so and it was 
planned to drop this for 2.9.0): my expectations are that we will 
support it in 2.9.0 and that USE_CHM will then control the options 
used to install the recommended packages, but that support may be 
withdrawn at some not too distant point.

I intend in the next week or so (GFF is Mar 20) to review what loose 
ends there are in package installation and Windows builds and tidy 
them up.  Once we have an alpha pre-release, reports on such matters 
will be welcome but until then they are a distraction.

-- 
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