[R] MASS need R >= 2.4.0!!!

Dirk Eddelbuettel edd at debian.org
Sun Oct 15 19:16:37 CEST 2006


On 15 October 2006 at 14:45, Ronaldo ReisJunior wrote:
| I have R 2.3.1 installed by debian páckage.
| 
| I install  only the base and recommended R packages from Debian source, all 
| others packages I install from the R source at CRAN.

We some 50 or more r-cran-* packages that you could apt-get install too...
 
| But, I try to use library MASS, but I received this message:
| 
| > library(MASS)
| Error: This is R 2.3.1, package 'MASS' needs >= 2.4.0
| 
| What is the problem?

A mistake I made in specifying too weak a constraint when I said 'Depends:
r-base-core (>> 2.3.1)' as 2.3.1-1 satisfies this.  So several CRAN et al
packages compiled using the different R 2.4.0 release candidates in Debian
unstable slipped into testing before R 2.4.0 itself did.

As discussed on r-sig-debian from where I quote from a post from two days ago:

-----------------------------------------------------------------------------
You can either

1) go to Debian unstable and install r-base-core, r-base-dev, r-mathlib, ...
   from the 2.4.0 release manually (or semi-automatically using the pinning
   feature of apt-get), or

2) go to snapshot.debian.net and get the previous version of the failing
   package (here r-cran-vr), or

3) use install.packages() or update.packages() within R to overwrite
   to package

4) wait for R 2.4.0 to hit testing.

I recommend the first choice, esp with apt-get pinning which makes it quite
easy.  Choice 2 is probably the least scary if 'pinning' is a bit
much. Choice 3 may be the easiest. Choice 4 obviously stinks.

Again, sorry about this.  I'll try hard not to do that again.
-----------------------------------------------------------------------------

The R 2.4.0 packages may actually enter testing today, so 4) may not be so
onerous.  They may get blocked by apparent conflict with rpy which may
require a manual push by the release team. I have contacted them.

Again, sorry -- this shouldn't have happened.  

Dirk


-- 
Hell, there are no rules here - we're trying to accomplish something. 
                                                  -- Thomas A. Edison



More information about the R-help mailing list