[Rd] R as a scripting engine

Prof Brian Ripley ripley at stats.ox.ac.uk
Sun Jan 11 21:18:33 CET 2009


Those of you tracking R development will have noticed that we are 
moving towards using R as a scripting engine.  (It is often overlooked 
as such.)  Thus far INSTALL, REMOVE, SHLIB and massage-examples have 
been moved to R.

Reasons:

- it is platform-independent and needs no other tools installed.
   No need to worry about strange 'sh' variants (as on AIX and in the
   past on Mac OS X) or to track Perl version changes and miniumun
   requirements.

- it is fast.  Shell scripts are notoriously slow[*] on Windows as
   launching a new process there is relatively much slower than on a
   Unix-alike, to the extent that we have had to use Perl scripts on
   Windows. The actual script of INSTALL in R runs twice as fast as the
   shell script on Unix or the Perl script on Windows.  (In part this
   comes about because R scripts can be pre-parsed if put in a package:
   the 1150 lines of INSTALL was taking a couple of seconds to parse.)

- it is much easier to maintain.  We only need one version, and any R
   user should be able to folllow it (and any member of R-core to
   maintain it).  And one version means only one set of the
   documentation and automatically keeping things in step.

- R has lots of useful features, including automatic support of
   (for files) different line endings and encodings.

Of course there is the one-time cost of doing the conversion, which 
has been triggered here by the need for new functionality.  It also 
means that there are minor changes to merge the Unix and Windows 
versions.  (Some of you will be hearing from Uwe Ligges about use of 
undocumented variables in Windows configure/Makevars.win, for 
example.)

In due course we see phasing out the use of Perl, at least at run time 
(and that might even be for 2.9.0).

[*] Running configure under Cygwin took over an hour on a fairly 
laoded Windows box earlier in the week.

-- 
Brian D. Ripley,                  ripley at 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



More information about the R-devel mailing list