discrepancy between PACKAGES file and contents of the contrib directory (PR#309)

[Herve Dreau]

Le mer, 03 nov 1999 08:53:58 Prof Brian D Ripleya écrit :
> Yesterday I asked privately which mirror's PACKAGES file this was after an
> earlier post, as I could not (and still cannot) get to Vienna. But AFAIK
> 

Sorry but I have had replied to you privatelly yesterday.

In this lost (?) mail i said :

 Thank you for your response.
 I use the master CRAN server, I notice some network weekness, but i think 
 if it is the master server all other are just mirror, and the PACKAGES file
 may be  the most recent in the master server.
 After your mail, i compare the content of the PACKAGES file in master CRAN
 and in some mirrors and i found that the tuwien one is the more accurate.
 
 The mirror in italy is particulary old (timestamp on the PACKAGES file is near one
 year old).
 The PACKAGES file in ethz.ch is nearly the same as the one in tuwien (minus the new (?)
  norm package).
 
 Her's a little (and quickly) summary of the discrepancy between the file PACKAGES and 
 the content  of the contrib directory in tuwien

... inserted the report send to r-bug


> rpart's entry had been updated after I reported that one
> rmtools is not on CRAN
> bats is in devel, not in the contrib directory
> ppr is in Old, not in the contrib directory
> nnet and spatial are nowhere (as part of VR), and VR is not a package
>   but a package bundle: we know those are not yet covered by
>   update.packages.

Perhaps but i think all the entry in PACKAGES are listed by 
CRAN.packages()
could not verify in this time : the url at tuwien is unreachable
but using another mirror :
CRAN.packages(CRAN="http://lib.stat.cmu.edu/R/CRAN")
[10,] "bats"                                 "0.1-2" 
[49,] "rmtools"                              "0.2"   
[43,] "ppr"                                  "1.1-2" 
[54,] "spatial"                              "5.3pl035-1"
[38,] "nnet"                                 "5.3pl035-1"
[43,] "ppr"                                  "1.1-2"     

It's only confusing to list packages that are unavailable and cause "false"
error message.

> 
> The update mechanism does depend on a consistent naming scheme, with
> DESCRIPTION entires and file names matching. Fritz Leisch is working on a
> checker, but we do rely on authors at present. For example. the DESCRIPTION
> file in splines_2.0-6.tar.gz has version 2.0-3. Authors: please do check!
> 
> 
> On Wed, 3 Nov 1999 hdreau@cybercable.fr wrote:
> 
> > Full_Name: Hervé Dréau
> > Version: not  applicable
> > OS: not  applicable
> 
> Sorry, those _are_ applicable. The Unix and Windows versions of
> update.packages are quite distinct. I have assumed Unix was meant.

Perhaps but it's not related to my version of R it's only a comparison
between the content of a file and the listing of a directory.
If that matter :
OS linux Redhat 6.1
R 0.65.1

Best regards

Hervé Dréau



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._