[R] Ignoring version numbers when installing packages...

[Duncan Murdoch]

On 30/03/2012 12:47 PM, Jonathan Greenberg wrote:
> Follow-up:
>
> r-forge seems to have some weirdnesses with version numbers, and since
> this is an incubation site, there are occasions where the version
> numbers were put in based on whatever the programmer was using at a
> time (e.g. other versions were simply not tested, so they erred on the
> safe side).  I understand that the internals may have changed, but I
> would argue that 9/10 times if something is working on R 2.14.1 but
> not 2.14.2 it has nothing to do with some incompatibility, just that
> the package hasn't been updated in some time and tested against the
> new version.
>
> So, to follow-up, the only way to "override" the version installation
> check is to modify the DESCRIPTION file?   There's no flag either in
> install.packages() or R CMD INSTALL that will force the installation
> even if the version check fails?

There are almost certainly other ways, but that seems simplest.  It's 
pretty unlikely that there would be an incompatibility between 2.14.1 
and 2.14.2, but if you try a 2.13.x package in 2.14.x, it's a much 
bigger risk.

Duncan Murdoch
>
> --j
>
> On Fri, Mar 30, 2012 at 11:41 AM, Duncan Murdoch
> <murdoch.duncan at gmail.com>  wrote:
> >  On 30/03/2012 12:05 PM, Jonathan Greenberg wrote:
> >>
> >>  R-helpers:
> >>
> >>  I'm trying to install a package from r-forge, and I'm wondering if
> >>  there is a way to force R to install a package, even if the package
> >>  "requires" a certain version of R (short of modifying the DESCRIPTION
> >>  file)?  Cheers!
> >
> >
> >  You don't give details, so the explanation is uncertain, but the answer is
> >  clear: don't do that.
> >
> >  If you are trying to install a binary build, don't do that.  It might appear
> >  to work but fail in messy ways if you use it in the wrong version of R.
> >    Some packages link to R internals, and those might have changed.
> >
> >  If you are trying to install from source, but the author said you need a
> >  particular version of R, then you could modify the DESCRIPTION file:  but
> >  presumably the author had a reason to ask for a particular version, so don't
> >  do that either.
> >
> >  Duncan Murdoch
> >
> >
>
>
>