[Rd] RFC install.packages() changes

[Liaw, Andy]

> From: Prof Brian Ripley
> 
> On Sun, 7 Nov 2004, Kjetil Brinchmann Halvorsen wrote:
> 
> > Prof Brian Ripley wrote:
> > 
> > >install/update.packages() in R-devel has been enhanced in 
> a few ways,
> > >notably to handle multiple repositories, and to allow 
> packages to be 
> > >reinstalled under the current version of R.
> > >
> > >1) Currently they will ask if you want to delete the 
> downloaded files (if 
> > >any) *provided* destdir was not specified.  In that case it uses a 
> > >subdirectory of the per-session temporary directory, so 
> the files will be 
> > >deleted at the end of the R session anyway.
> > >
> > >We had a request to make the questioning optional.  I 
> propose to remove it 
> > >altogether, and either
> > >
> > >a) delete the downloaded files at the end of install.packages or
> > >b) leave them in the per-session directory to be deleted 
> at the end of the 
> > >session.
> > >
> > >that is to force answer 'y' or 'n' respectively.  Does 
> anyone see any 
> > >problems with a)?
> > >
> > >  
> > >
> > I would prefer b), as I am downloading from the Rgui menu 
> and then moves 
> > downloaded packages
> > to a c:\download... area which I use to burn CD-roms I 
> distribute to 
> > many people. If  a) is choosen I guess
> > I could learn to use it from the command line, though. What is the 
> > problem with b)?
> 
> Potentially leaving 1Gb or so of files in a temporary directory. So 
> probably no big deal.

My personal preference would be a).  Kjetil's usage seems like a bit of an
abuse, and one should be able to write something that does that specific
task (e.g., modify update.packages() to something called
download.packages()), rather than relying on a function in R that's meant
for something else.  

One thing that concerns me a bit (I haven't check whether the concern is
warranted) is package dependency:  If the new version of a package on CRAN
requires, e.g., a newer version of R than what the user has, what happens
then?

Andy

 
> Brian
> 
> -- 
> 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
> 
> ______________________________________________
> R-devel at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
> 
>