On Thu, 7 Oct 2004, Petr Pikal wrote:
> Hi all
>
> I upgraded to 2.0.0 version and did everything as I used to do
> before.
>
> I installed windows binary, copy/paste other than bundled
> packages.
>
> I got e.g.
> > library(chron)
> Error in library(chron) : 'chron' is not a valid package --
installed <
> 2.0.0?
>
> so I loaded it from CRAN and everything worked OK except my
> own personal functions (they are not on CRAN). So I went
> through NEWS file which says:
>
> o Packages must have been re-installed for this version, and
> library() will enforce this.
>
> I have never done it, my simple set of functions was never actually
> installed. I used copy/paste the directory from old R version to
> new one and I used library(fun) in personal .Rprofile to load my
> function set. Doing that again I got:
>
> Error in library(fun) : There is no package called 'fun'
>
> I know I can use source
> source("D:/programy/R/rw2000/library/fun/R/fun")
> but I feel library way is better way and I think I could try it.
>
> Before I start to try transferring my function set to a real package I
> would like to ask simple ***question***.
>
> Is the error message result of not installing package fun correctly
> (at all) or I shall do something more than follow instructions in
> README.packages and other documentation?
I believe that message comes from your self-built package not having a
DESCRIPTION file.
Installing a package checks it has the features needed (such as a
DESCRIPTION file) and then creates a lot of metadata (indices, parsed
NAMESPACE and DESCRIPTION files and so on) that R assumes will be present
when a package is loaded. So as of 2.0.0 you do need to use R CMD INSTALL
to create a usable package.
--
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