[R-sig-DB] Re: Rdbi package [forwarded msg]

[David James]

Hi,
  
We did start an informal discussion back in April, right after Tim
submitted his proposal.  I think that we agreed that writing some
kind of draft "request for comments" (RFC) based on his initial Rdbi
was a good idea.  At the time, Brian correctly pointed out that we
had no framework for discussion, and that there were many other
pressing issues that came up at the Vienna conference requiring
attention for the members of the r-core group.  Since then, a
lot of good things have happened: Martin has setup the R-SIG-DB,
and versions 1.3.0 and 1.3.1 have been released (among other nice
things).  I'd agree with Kurt that it's probably a good time to
get back to Tim's proposal.

Is there a consensus that an RFC based on the Rdbi is the next
logical step?  

Martin Maechler wrote:
> This definitely belongs to the SIG (special interest group) mailing list,
> as well {to which probably all in "CC" are subscribed, 
> 	 but then I want to be sure ..}
> BTW, as a reminder : http://www.stat.math.ethz.ch/mailman/listinfo/r-sig-db
> 
> Martin
> 
> ------- start of forwarded message -------
>  From: Kurt Hornik <Kurt.Hornik using ci.tuwien.ac.at>
>  To: R-core using r-project.org, "Timothy H. Keitt" <Timothy.Keitt using stonybrook.edu>,
> 	 David James <dj using research.bell-labs.com>,
> 	 Saikat DebRoy <saikat using stat.wisc.edu>,
> 	 Torsten Hothorn <Torsten.Hothorn using rzmail.uni-erlangen.de>
>  Subject: Rdbi package
>  Date: Tue, 4 Sep 2001 22:54:53 +0200
> 
> Tim Keitt has created a new package called `Rdbi', which implements
> `a generic framework for database access in R', plus another packaeg
> called `Rdbi.PgSQL' which implements a PostgreSQL driver using the
> generic tools, and is a complete rewrite of the former RPgSQL package.
> 
> As far as I understand it, Rdbi uses ideas from the RS-DBI draft, but
> does not provide an implementation of it.  My understanding/hope based
> on discussions in Vienna was that we would have a working group (SIG)
> which would improve the standard, and maybe also provide code that works
> under all implementations of the S language.  Tim only uses R, and it is
> not clear whether his code works under Splus (hence also `Rdbi' versus
> `RSdbi').
> 
> As the new package might very well constitute a de-facto standard, I
> have put the packages into the CRAN contrib/Devel section for the time
> being, so that in particular the maintainers of the other DBMS access
> packages can have a look, and maybe can give some comments.
> 
> Waiting for feedback and instructions on how to proceed further ...
> 
> -k
> ------- end of forwarded message -------

-- 
David A. James
Statistics Research, Room 2C-253            Phone:  (908) 582-3082       
Bell Labs, Lucent Technologies              Fax:    (908) 582-3340
Murray Hill, NJ 09794-0636