[R] Re: [R-sig-finance] R for CGI

[Dirk Eddelbuettel]

On 29 January 2005 at 10:35, Andrew Piskorski wrote:
| On Fri, Jan 28, 2005 at 12:30:53PM -0800, ebashi wrote:
| 
| > Perl is the common language to write CGI scripts which handle
| > Forms. My question is that can R be as fast as perl to do the same
| > job(with using CGIwithR package). Is it an optimal solution to
| > connect R directly to a commercial HTML webpages,
| 
| First of all, why are you asking this on the r-sig-finance list?  The
| question does not belong there.

Yup, and your's truly, with his r-sig-finance listmaster hat on, has
subsequently unsubscribed Mr "ebashi" from r-sig-finance as he has 

 * repeatedly crossposted (despite strong hints that this is not looked upon
   too kindly),
 * repeatedly asked essentially the same question, and nevertheless 
 * continues to persistently ignore to good advice given to him.

Going e.g. to this listarchive which can sort by author, you see for January
(URL is http://tolstoy.newcastle.edu.au/R/help/05/01/author.html)

  # ebashi

      * [R] R for CGI (29 Jan 2005)
      * [R] How to make R faster? (26 Jan 2005)
      * [R] CGIwithR (17 Jan 2005)

along with another R/PHP post in December (URL
http://tolstoy.newcastle.edu.au/R/help/04/12/author.html) 

  # ebashi

      * [R] R&PHP (29 Dec 2004)
      * [R] xy_plot (01 Dec 2004)

If you read those threads -- which the original poster clearly must have
avoided at almost all cost -- you'll find many good points, answers and tips
for further references.

But no, "ebashi" rather asks the same question over and over and over. For
good measure, I also got once personally to my inbox.

| Secondly, if you care about speed and "optimal solutions", CGI is
| absolutely the last thing you want to use, regardless of whether you
| write your scripts in Perl, R, or any other language.
| 
| For high-performance dnamic web pages, the most typical approach is to
| embed a scripting language interpretor directly into the web server -
| Tcl for AOLserver, mod_perl for Apache, etc.  Alternative approaches
| include designs like FastCGI.  This is basic web stuff that was all
| figured out long ago, perhaps c. 1997.  (Go google and read up on it.)
| 
| I don't have any links handy, but there are definitely some existing R
| projects which let a web server efficiently evaluate R code by passing
| it to an already running R process/server.  That's what most people
| want, not to build a sophisticated dynamic website entirely in R.
| 
| On the general web front, here are some ancient (but good)
| introductions to some of the basic concepts:
| 
|   http://philip.greenspun.com/panda/
|   http://philip.greenspun.com/wtr/aolserver/introduction-1.html
| 
| And the docs for one good current toolkit which uses all those ideas:
| 
|   http://openacs.org/doc/

All very good points.

Dirk

-- 
Better to have an approximate answer to the right question than a precise 
answer to the wrong question.  --  John Tukey as quoted by John Chambers