R help - Apr 2002 - R package organization

>6)  Better layout of packages listed on CRAN.  This listing
format will
>collapse under its own weight once it gets too large.
>
As a new R user, I have thought for some time that the organization
of packages, and functions within packages, in R needs some more
coherent structure, perhaps something like the hierarchy of classes
used in Perl and contributed to CPAN.  For example, Perl modules
are listed on CPAN organized in a tree containing top-level categories
like

Archiving and Compression
Commercial Software Interfaces
Control Flow Utilities
Data Type Utilities
Database Interfaces
Development Support
File Handle Input Output
  ...
each of these contain numerous subcategories, and there is a
coherent relationship between the subcategories and the directory
structure for installed modules.  So, to use functions to search
for files on my file system, I say

use File::Find;

and if someone writes a new module, say FastCopy.pm it goes into
/usr/lib/perl5/File/FastCopy.pm in my Perl library.

In R, on the other hand, everything goes into R/rw104x/library,
and each pachage contains all its functions within packagename/R,
so there is only a very flat organization across packages
and within packages.

Some packages are quite coherent and relate to a single general
class of methods (John Fox's sem package, for example), while
others are a heterogeneous collection of tools which someone
found useful, and kindly took the effort to document and make public.
 
This makes it quite difficult, sometimes to find a function for
a given task, and I see this reflected both in my own
efforts to use R as well as in many queries on this list, often
answered repeated with a brief

library(foo)
?bar

It also makes it difficult for one person to extend another's
work, e.g., by writing new functions or enhancing others
from an existing package.

This is not a criticism of the R Development Team, but, as
R continues to grow, and more people make contributions to
CRAN, perhaps it is time for some thought to be given to a
reorganization of package structure.

As I look over the list of packages on CRAN, I don't see
any obvious categories, but perhaps others will have some
ideas on this topic.

-Michael

-- 
Michael Friendly              friendly at yorku.ca
York University               http://www.math.yorku.ca/SCS/friendly.html
Psychology Department
4700 Keele Street             Tel:  (416) 736-5115 x66249
Toronto, Ontario, M3J 1P3     Fax:  (416) 736-5814
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On 04/03/02 09:44, Michael Friendly wrote:
>
>>6)  Better layout of packages listed on CRAN.  This listing
>format will
>>collapse under its own weight once it gets too large.
>>
>
>As a new R user, I have thought for some time that the organization
>of packages, and functions within packages, in R needs some more
>coherent structure, perhaps something like the hierarchy of classes
>used in Perl and contributed to CPAN.  For example, Perl modules
>are listed on CPAN organized in a tree containing top-level categories
>like [...]
As an occasional CPAN user, I never use this and doubt I ever
will.  I usually go there for some specific thing that I hear
about somewhere else or find through Google (e.g., html2latex).

I tend to think that the effort involved in making a good
classification would be better spent in making better search
tools, or something else altogether.  I also think that CRAN may
require a lot more expertise to understand than CPAN does.

But there is a simple way to settle this.  You, or anyone, could
copy the CRAN page and re-organize it.  Announce it, and ask for
comments.  See if it flies.

Jon Baron
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Wednesday 3 April 2002 16:44, Michael Friendly wrote:
> As I look over the list of packages on CRAN, I don't see
> any obvious categories, but perhaps others will have some
> ideas on this topic.
It does not have to be just one map... Freshmeat.net and some other
sites use something called Trove categories (e.g. 
http://freshmeat.net/projects/jchempaint).
There can be several trees in which a package can be found, possibly even
at different leaves in the same tree.

Example trees for R could based on general method or data type, or whatever.
A tree could be based on chapters in some standard encyclopedia, for example.
We could develop several trees, each one aiming at a specific discipline: one 
for chemometrics, one for sociometrics, econometrics etc...

A paradigma in webdesign is that each webpage should be accessible within at 
most three click, otherwise it will not be read... the same holds for these 
kind of search trees (compare for example with descision trees, where the 
depth is rather irrelevant)...

If adding more trees make packages more used, we gain something...

Egon


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help 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-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._