This is really an R-devel question -- it is about an unreleased version of
R.
On Thu, 5 Aug 2004, Jeff Gentry wrote:
> Hi there ...
>
> Not too long after the switch to using Subversion I tried to checkout &
> build R but encountered an error - and being short on time at that point
> put it off to look at later. So today I sat down again and was
> encountering this error every time I would attempt to build:
>
> dumping R code in package 'methods'
> Saving namespace image ...
> initializing class and method definitions now ...done
> <environment: namespace:methods>
> usage: touch [-acfm] [-r file] [-t [[CC]YY]MMDDhhmm[.SS]] file ...
> *** Error code 1
>
> I traced this down to the Makefile in methods:
> $(top_builddir)/library/$(pkg)/R/$(pkg).rdb:
$(top_builddir)/library/$(pkg)/R/all.rda
> @echo "tools:::makeLazyLoading(\"$(pkg)\")" | \
> R_DEFAULT_PACKAGES=NULL LC_COLLATE=C $(R_EXE) --slave >
> /dev/null
> @cat /dev/null > $(top_builddir)/library/$(pkg)/R/all.rda
> @touch $^
>
>
> I'm not sure what it is about that touch call, but commenting it out
> causes things to build normally. When I try to echo '$^', it
appears to
> be an empty string.
>
> This is all on a FreeBSD machine - I'm assuming it is somehow OS
specific
> as this would have come up a bunch of times if it wasn't.
Yes. Seems a make peculiarity, not even shared by Solaris make.
To get you going, replace $^ by
$(top_builddir)/library/$(pkg)/R/$(pkg).rdb. However, I am working right
now on streamlining this now we don't allow lazy-loading to be optional.
--
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