R devel - Jan 2005 - Struggling with S3/S4 interface issues and External Pointers

I'm currently working on embedding R into the Apache2 webserver (with 
some positive results...), but I'm struggling to identify the best way 
to expose the apache data and functions to R.

A couple of thoughts:

In light of the recent discussion on "S3/S4 classes performance 
comparson", I'm leaning toward implementing the interface with S3 style
classes. I appreciate anyones advice on this.

I JUST found (after roughly 6 months of studying R) Luke Tierney's note 
on "Finalization and Weak References in R". From the developers page, 
this note is listed as a "preliminary mechanism", but I scoured CRAN
and
noticed packages like RODBC and XML are using it. Has this mechanism 
changed from "preliminary" to "production" yet? For storing
pointers in
R variables, I've been using integer vectors which is what RMySQL does 
but doesn't seem natural.

For those interested in my Apache R implementation, please download here:

http://biostat.mc.vanderbilt.edu/twiki/pub/Main/JeffreyHorner/mod_R-0.0.tar.gz

Cheers,
-- 
Jeffrey Horner       Computer Systems Analyst         School of Medicine
615-322-8606         Department of Biostatistics   Vanderbilt University

On Fri, 21 Jan 2005, Jeffrey Horner wrote:
> I'm currently working on embedding R into the Apache2 webserver (with
some
> positive results...), but I'm struggling to identify the best way to
expose
> the apache data and functions to R.
>
> A couple of thoughts:
>
> In light of the recent discussion on "S3/S4 classes performance
comparson",
> I'm leaning toward implementing the interface with S3 style classes. I 
> appreciate anyones advice on this.
> I JUST found (after roughly 6 months of studying R) Luke Tierney's note
on
> "Finalization and Weak References in R". From the developers
page, this note
> is listed as a "preliminary mechanism", but I scoured CRAN and
noticed
> packages like RODBC and XML are using it. Has this mechanism changed from 
> "preliminary" to "production" yet? For storing pointers
in R variables, I've
Yes, but the documentation has not.  R itself uses the mechanism for 
DLLinfo.

The developer.r-project.org site is a mixture of old and new, but 
certainly worth perusing in the first month of studying R.
> been using integer vectors which is what RMySQL does but doesn't seem 
> natural.
You asked elsewhere why RMySQL does not use it, but AFAIK it predates the 
mechanism.  RODBC and XML have only used it recently.

-- 
Brian D. Ripley,                  ripley@stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

Prof Brian Ripley wrote:
[...]>
> You asked elsewhere why RMySQL does not use it, but AFAIK it predates 
> the mechanism.  RODBC and XML have only used it recently.
Actually, Luke suggested it be used when DBI was just forming:

https://stat.ethz.ch/pipermail/r-sig-db/2001-October/000028.html

-- 
Jeffrey Horner       Computer Systems Analyst         School of Medicine
615-322-8606         Department of Biostatistics   Vanderbilt University